MATERNAL AND FETAL REPRESENTATIONS, DIMENSIONS OF PERSONALITY, AND PRENATAL ATTACHMENT IN WOMEN HOSPITALIZED WITH HIGH RISK PREGNANCY APPROVED BY SUPERVISORY COMMITTEE Chair: H. M. Evans, Ph.D. Wayne H. Denton, M.D., Ph.D. Sandra Pitts, Ph.D. Richard Robinson, Ph.D. C. Allen Stringer, M.D.
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
MATERNAL AND FETAL REPRESENTATIONS, DIMENSIONS OF
PERSONALITY, AND PRENATAL ATTACHMENT IN WOMEN HOSPITALIZED
WITH HIGH RISK PREGNANCY
APPROVED BY SUPERVISORY COMMITTEE
Chair: H. M. Evans, Ph.D.
Wayne H. Denton, M.D., Ph.D.
Sandra Pitts, Ph.D.
Richard Robinson, Ph.D.
C. Allen Stringer, M.D.
DEDICATION
To mothers and babies everywhere, including mine,
and to David.
MATERNAL AND FETAL REPRESENTATIONS, DIMENSIONS OF PERSONALITY,
AND PRENATAL ATTACHMENT IN WOMEN HOSPITALIZED
WITH HIGH RISK PREGNANCY
by
ANNA RACHEL BRANDON
DISSERTATION
Presented to the Faculty of the Graduate School of Biomedical Sciences
The University of Texas Southwestern Medical Center at Dallas
In Partial Fulfillment of the Requirements
For the Degree of
DOCTOR OF PHILOSOPHY
The University of Texas Southwestern Medical Center at Dallas
Dallas, Texas
June 2006
Copyright
by
Anna Rachel Brandon, 2006
All Rights Reserved
ACKNOWLEDGMENTS
Dr. Monty Evans told me the idea for research at Baylor on attachment was
conceived in a conversation he and Dr. Allen Stringer had after an inspiring presentation by
Dr. Linda Mayes a few years ago. When it became time for them to implement the project, I
happened to be a graduate student in the right place at the right time. I am ever grateful for
the trust they had in me, and for their continued support throughout the project.
I approached the dissertation process with the traditional angst, but each member of
this committee has done their utmost to help me through it. Together, Dr. Evans and Dr.
Sandy Pitts have been my “secure base,” continually encouraging me to stretch academically
and clinically. Dr. Evans’ office door was always open to me, and his confidence in me took
over when my own confidence failed. Dr. Pitts poured over articles to help me develop my
research questions, and tirelessly scored ORI’s so we would have two raters. She has also
modeled for me an amazing ability to laugh, even in the most stressful times. Dr. Richard
Robinson generously plowed into a project already in action, helped us make the necessary
corrections, and patiently guided me through the statistical mazes I had been dreading. Dr.
Wayne Denton was always understanding and ever encouraging, and never minded my
bouncing between his important Couples project and my work at Baylor. Dr. Stringer not
only made available the Baylor population for our work, but made possible my additional
training at Yale to ensure a high-quality project. I cannot conceive of any better team of
advocates and collaborators.
My classmates, Paula Miltenberger and Dana Broussard, have made significant
contributions to this work as well. Paula took the lead as the work was initiated here at
Baylor, and continued to invest in the project even while on bedrest during her own
complicated pregnancy. Dana joined us in the midst of chaos, rolled up her sleeves, and was
ever alert to new ways to streamline our operations or minimize data loss. We three could
never have kept up with the volume of work without three amazing volunteers. Daria Dato
was a key team member as we began planning the project, and continued to work through our
first six months of data collection, always willing to take on any task we needed done. Missy
Heusinger and Georgina Rangel also played valuable roles in the data collection and day-to-
day needs of our research. I am grateful to all of them.
On a personal front, my husband David has become the neighborhood hero. He did
the laundry, the shopping, the housework, and guarded my time with vigilance. David
seemed to always know when I could work a little harder, and when I needed to stop. He
believed I could do this work, and he placed its importance over his own needs. If a
dissertation could have a “producer,” he would be mine. All five of our children have
supported my continued education, even though it meant I could not always be there for them
in the way I would have liked to be. And my three beautiful grandchildren are too young to
know what this is that has kept me cloistered in the back of our home for so long, but they’ve
been there to lift my spirits or make me take a break.
Aside from the valuable education I have received, my life is richer for knowing and
working with all of these people.
MATERNAL AND FETAL REPRESENTATIONS, DIMENSIONS OF PERSONALITY,
AND PRENATAL ATTACHMENT IN WOMEN HOSPITALIZED
WITH HIGH RISK PREGNANCY
Publication No.
Anna Rachel Brandon, Ph.D.
The University of Texas Southwestern Medical Center at Dallas, 2006
H. M. Evans, Ph.D.
The present study investigated the effects of self-criticism, dependency, object representation, and risk upon maternal antenatal attachment in women hospitalized during pregnancy with high risk of maternal or fetal demise. Ninety-one women completed the Depressive Experiences Questionnaire (Blatt, D’Affliti, and Quinlan, 1976), the Object Relations Inventory (Blatt et al.,1992), the Maternal Antenatal Attachment Scale (Condon, 1973), the Edinburgh Postpartum Depression Scale (Cox, Holden, and Sagovsky, 1987) and the Center for Epidemiological Studies Depression Scale (Unauthored, 1999) within the first three days of hospital admission. No relationship was indicated between maternal representations and antenatal fetal attachment, nor was there a correlation between maternal representation and fetal representation. Self-critical mothers significantly scored lower in the measure of antenatal attachment quality and endorsed a higher number of depressive symptoms. Mothers hospitalized because of maternal risk were not significantly different in their reports of attachment than were mothers hospitalized because of fetal risk, and no significant differences were found across severity of risk factors as evaluated by the Hobel Risk Assessment. Consistent with previous research, depressive symptomatology was associated with a lower quality of maternal antenatal attachment overall. Results suggest that maternal narratives may not be significantly linked with reported antenatal attachment and depressive symptoms have a stronger association with reductions of antenatal attachment than dependent or self-critical tendencies.
vii
TABLE OF CONTENTS List of Definitions 2 Chapter One: Introduction 4 History of Attachment Theory 4 The Conceptualization of Prenatal Attachment 8 The Measure of Prenatal Attachment 12 Criticism of the MFA Construct 16 The Relevance of Prenatal Attachment 18 Purpose of the Study 23 Chapter Two: Review of the Literature 25 Search Methods 26 Internal Working Models 26
Background 26 Integration of Theory 29 MFA and Object Representation 31
The Contribution of Personality 39 The Impact of Risk 46 Conclusion of Review 54 Rationale 54 Aims 56
Hypotheses 56 Primary Hypotheses 56 Secondary Hypotheses 57 Chapter Three: Methodology 59 Participants 59 Methods and Procedures 59 Measures 60 Chapter Four: Results 66 Demographic Information 66 Overview of Statistical Analyses 68 Chapter Five: Conclusions and Recommendations 84 Characteristics of the Sample 85 Discussion of Findings 89 Theoretical Implications 100 Clinical Implications 100 Limitations and Future Directions 102 Conclusion 105
viii
PRIOR PUBLICATIONS
Brandon, A. R.; Pitts, S.; Robinson, R.; Stringer, C.A. (2006). “Preliminary findings on the
associations of object representation and personality with prenatal attachment during high-
risk pregnancy.” Presented at the Winter 2006 Meeting of the American Psychoanalytic
Association, New York City.
Denton, W. & Brandon, A. R. (2006). Couple therapy in presence of mental disorders. Journal of
Couple and Relationship Therapy, In Press.
McCullough, M. M.; Orsulak, P.; Brandon, A.; & Akers, L. (2005). Rumination, fear and cortisol:
An in vivo study of interpersonal transgressions. Health Psychology, In Press.
ix
LIST OF FIGURES
Figure 1 108 Levels of Mother Narrative Conceptual Level Across Low, Medium, and High Quality of Attachment Scores
Figure 2 109
Levels of Mother Narrative Conceptual Level Across Low, Medium, and High Intensity of Attachment Scores
Figure 3 110
Quality of Attachment and Mother Narrative Conceptual Level Figure 4 111
Intensity of Attachment and Mother Narrative Conceptual Level Figure 5 112
Global Attachment and Mother Narrative Conceptual Level Figure 6 113
Distribution of Levels of ORI Narrative Conceptual Level Figure 7 114
Attachment Style Based on Below and Above Means of MAAS Quality and Intensity Factors Figure 8 115
Distributions of ORI Conceptual Levels of Mother Narrative Across DEQ Self-Criticism Scores Figure 9 116
Distributions of ORI Conceptual levels of Mother Narrative Across DEQ Dependency Scores Figure 10 117 Type of Risk and MAAS Global Attachment Scores Figure 11 118 Type of Risk and MAAS Intensity of Attachment
x
LIST OF TABLES
Table One 119 Demographic Characteristics of Total Sample Table Two 122 Pregnancy Characteristics of Sample Table Three 123 Psychiatric Characteristics of Sample Table Four 124 Means and Standard Deviations of Measures Table Five 126 Means and Standard Deviations of the Object Relations Inventory Mother Narrative for Global Attachment Above and Below the Mean Table Six 127 Means and Standard Deviations of the Object Relations Inventory Mother Narrative for Global Attachment by Standard Deviation Table Seven 128 95% Confidence Intervals of Pairwise Differences in Mean Changes Of Attachment Quality (MAAS) by Conceptual Level of ORI Mother Narrative Table Eight 129 95% Confidence Intervals of Pairwise Differences in Mean Changes Of Attachment Intensity (MAAS) by Conceptual Level of ORI Mother Narrative Table Nine 130 Frequency Distribution of Conceptual Levels of Mother and Baby Narrative Table Ten 131 Spearman’s rho Correlations for Conceptual Level and Attachment Table Eleven 132 Two-Way Contingency Table of Levels of Object Representation Of Mother and Quality of Attachment Above and Below the Mean
xi
Table Twelve 133 Two-Way Contingency Table of Levels of Object Representation Of Mother and Quality of Attachment by Standard Deviation Table Thirteen 134 Two-Way Contingency Table of Levels of Object Representation Of Mother and Intensity of Attachment Above and Below the Mean Table Fourteen 135 Two-Way Contingency Table of Levels of Object Representation Of Mother and Intensity of Attachment by Standard Deviation Table Fifteen 136 Means and Standard Deviations on Dimensions of the Object Relations Inventory for Global Attachment Above and Below the Mean Table Sixteen 137 Means and Standard Deviations on Dimensions of the Object Relations Inventory for Global Attachment by Standard Deviation Table Seventeen 138 Spearman’s rho Correlations for Object Representations of Mother and Baby Narratives Table Eighteen 139 Two-Way Contingency Table of ORI Conceptual Level of Mother Narrative by Baby Narrative Table Nineteen 140 Pearson Product-Moment Correlations of Dependency, Self-Criticism, and Maternal Antenatal Attachment Table Twenty 141 Two-Way Contingency Table of Self-Criticism (Low, Average, and High) and Antenatal Attachment Style Table Twenty-one 142 Two-Way Contingency Table of Self-Criticism (Above and Below the Mean) and Antenatal Attachment Style Table Twenty-two ` 143 Two-Way Contingency Table of Dependency (Low, Average, and High) and Antenatal Attachment Style
xii
Table Twenty-three 144 Two-Way Contingency Table of Dependency (Above and Below the Mean) and Antenatal Attachment Style Table Twenty-four 145 Two-Way Contingency Table of Self-Criticism (Low, Average, and High) and Conceptual Level of Mother Narrative Table Twenty-five 146 Two-Way Contingency Table of Dependency (Low, Average, and High) and Conceptual Level of Mother Narrative Table Twenty-six 147 Two-Way Contingency Table of Object Representation and Self- Critical Characteristics (Below the Mean and Above the Mean) Table Twenty-seven 148 Two-Way Contingency Table of Object Representation and Dependent Characteristics (Below the Mean and Above the Mean) Table Twenty-eight 149 Hobel Risk Assessment Factors Identified as Fetal Table Twenty-nine 150 Hobel Risk Assessment Factors Identified as Maternal Table Thirty 151 Means and Standard Deviations of MAAS Global Antenatal Attachment Scores Across Three Types of Risk Table Thirty-one 152 Means and Standard Deviations of MAAS Global Antenatal Attachment Scores Across Two Types of Risk Table Thirty-two 153 95% Confidence Intervals of Pairwise Differences in Mean Changes MAAS Attachment Intensity by Type of Risk Table Thirty-three 154 Two-Way Contingency Table of Three Risk Types and Antenatal Attachment (MAAS)
xiii
Table Thirty-four 155 Two-Way Contingency Table of Three Risk Types and Maternal Antenatal Attachment Table Thirty-five 156 Pearson Product-Moment Correlations for Severity of Risk and Maternal Antenatal Attachment Table Thirty-six 157 95% Confidence Intervals of Pairwise Differences in Mean Changes MAAS Attachment Intensity by Level of Risk Table Thirty-seven 158 Pearson Product-Moment Correlations for Gestational Age and Antenatal Attachment Table Thirty-eight 159 Pearson Product-Moment Correlations for Depressive Symptoms, Attachment, and Risk Table Thirty-nine 160 Correlations Between Major Demographic Variables, Depressive Symptoms (EPDS), and Antenatal Attachment (MAAS) Table Forty 161 95% Confidence Intervals of Pairwise Differences in Means of EPDS (Depressive Symptoms) Table Forty-one 162 Chi-Square Comparison of EPDS Depressive Symptomatology and MAAS Antenatal Attachment Style Table Forty-two 163 Sample Means and Standard Deviations of Standard and Three Alternate Versions of Scoring of the Depressive Experiences Questionnaire Table Forty-three 164 Sample Intercorrelations of Four Scoring Methods of the Depressive Experiences Questionnaire
xiv
Table Forty-four 165 Pearson Product-Moment Correlations of DEQ Dependency, DEQ Self-Criticism, and MAAS Antenatal Attachment Using the McGill Scoring Method Table Forty-five 166 Pearson Product-Moment Correlations of DEQ Dependency and Relatedness with MAAS Antenatal Attachment Table Forty-six 167 Pearson Product-Moment Correlations of MAAS Antenatal Attachment And Rude & Burnham’s Needy and Connectedness DEQ Scoring Method Table Forty-seven 168 Linear Regression Analyses of Dependency and Self-Criticism Scores Predicting MAAS Global Attachment Score Table Forty-eight 169 Spearman’s rho Correlation of ORI Baby Narrative and Gestational Age (Weeks) Table Forty-nine 170 Comparison of Most Common Risk Factors Table Fifty 171 Comparison of the ORI and MAAS Means and Standard Deviations from Two Samples Table Fifty-one 172 Comparisons of Correlations for ORI Mother Narrative and Maternal Antenatal Attachment in Two Samples
LIST OF DEFINITIONS Antenatal—period of gestation also referred to as “prenatal” and “antepartum.”
Antepartum—period of gestation, also referred to as “antenatal” and “prenatal.” Antepartum depression—in this discussion, refers to the onset of a major depressive episode or minor depression during pregnancy. External validity—the extent to which a study’s conclusions can be applied to populations and settings outside those of the study itself. Incidence—the percentage of the population with an illness episode that begins within a given period of time (e.g., during pregnancy or within the first 3 months following delivery). Internal Working Model (IWM)—dynamic mental representations or “templates” constructed by infants of their interpersonal world that shape expectations, responses, and interpretations of interpersonal behavior Major depressive disorder—a type of mood disorder characterized by one or more major depressive episodes. The Diagnostic and Statistical Manual, version IV, Text Revision, (DSM-IV-TR) defines a major depressive episode as a period of at least 2 weeks during which an individual experiences daily disturbance in mood (intense feelings of sadness), or loss of interest in activities that have been pleasurable in the past, and at least four of eight symptoms: (1) hypersomnia or hyposomnia, (2) changes in appetite or loss/gain of weight, (3) psychomotor agitation or retardation, (4) loss of energy (fatigue), (5) feelings of worthlessness or excessive guilt, (6) problems with concentration, (7) loss of interest in sex, and (8) recurrent suicidal thoughts or suicidal attempt. These symptoms must be present most of the day and nearly every day during the 2-week episode, must cause clinically significant distress or impairment in functioning, and must not be the result of the direct physiologic effects of a substance or a general medical condition. Major depressive disorder is not diagnosed if symptoms are attributable to an acute grief reaction; however, it is diagnosed after an acute grief reaction if the syndrome persists for two months or longer. It is not diagnosed if there is a history of manic, hypomanic, mixed episodes, or schizophrenia. Minor depressive disorder—a subthreshold diagnosis with a number of definitions (also referred to as minor depression). This term usually describes one or more episodes of depression lasting two weeks or longer but with fewer symptoms than required for major depressive disorder diagnosis. Multigravida--a pregnant woman who has carried a previous fetus to viability, regardless of outcome.
1
Multipara—a woman who has carried more than one fetus to viability, regardless of whether the offspring were born alive. Multiparity is the condition of having carried one or more fetuses to viability, and multiparous describes a woman who has borne more than one child. Object Representation—A psychoanalytic term referring to the internal mental representational aspects of a significant other person, incorporating both realistic and fantasied characteristics of the relationship. Perinatal—during pregnancy (also referred to as prenatal) and 12 months following delivery. Period Prevalence—the percentage of the population with depression over a specific period of time (e.g., during pregnancy or from delivery to the end of the first six weeks, the first three months, or the first year, etc.). Postnatal—period of time following birth referring to the infant Postpartum—period of time following birth referring to the mother lasting from parturition to 12 months after delivery. Postpartum depression—the DSM-IV-TR defines this as a specific type of major depressive disorder that has an onset of a major depressive episode within 4 weeks after delivery. In this discussion, the term will be expanded to include minor depression. Point Prevalence—the percentage of the population with depression at a given point in time Prenatal—the period of pregnancy from conception to parturition. Primagravida—A woman during her first pregnancy. Primapara—A woman who has been delivered of one infant of at least 20 weeks gestation regardless of its viability, and primaparous describes a woman in the period of time surrounding her first pregnancy. Puerperium—the 6-week period following delivery. Reliability—the extent to which a test, inventory, or scale is consistent in its evaluation of the same individuals Screening instrument—a measure or test utilized to identify an individual with respect to likelihood of having a specific disorder. A screen itself does not provide a diagnosis, however, when positive, indicates that further investigation is necessary to confirm (or disconfirm) the presence of the disorder.
2
Sensitivity—the ability of a measure or test to correctly identify those with a syndrome, calculated as the percentage of true positive values compared to false negative values. Specificity—the ability of a measure or test to correctly identify those who do not have a syndrome, calculated as the percentage of true negative values compared to false positive values. Validity—the extent to which a test, inventory, or scale measures what it is supposed to measure.
3
CHAPTER ONE Introduction
HISTORY OF ATTACHMENT THEORY
John Bowlby, a young volunteer at a school for maladjusted children, was moved by
his experiences with two young boys—one isolated and distant, the other anxious and
clinging (Ainsworth, 1968). Since neither child had a stable mother figure, he wondered if
early family relationships had profound effects upon the personality development of children.
His curiosity and desire to explore this idea, coupled with his rigorous scientific training at
the University of Cambridge, led to a reevaluation of career goals and the decision to become
a child psychiatrist (Bretherton, 1992; Senn, 1977). Subsequent work led to his formulation
of the basic tenets of what is known today as “attachment theory,” a synthesis of elements
from ethology, cybernetics, information processing, developmental psychology, and
psychoanalysis. Bowlby’s original work focused on the infant’s biological need for a
secure early attachment to the mother and the mother’s response, a major conclusion being
that a maturing child’s mental health fundamentally required that “the infant and young child
should experience a warm, intimate, and continuous relationship with his mother (or
permanent mother substitute) in which both find satisfaction and enjoyment” (Fonagy,
2001a; Bowlby, 1951; Bowlby, 1969).
Bowlby conceptualized the attachment system as an evolutionary set of behaviors
related to those of exploration, fear, affection, and caregiving. Regulation of this system,
Bowlby reasoned, was solely biological; he posited that the infant’s primary goal was to
maintain a certain degree of physical proximity to the mother for survival. Bowlby later 4
added to his stance that attachment would include psychological goals on the part of the
developing child and mother, but his insistence that attachment was an independent
behavioral system and not related to unconscious drives was a solid wedge between his
theory and the psychoanalytic theories of his training. Even though this assumption
separated him from the analytic community, key researchers such as Mary Ainsworth, James
Robertson, Rudolph Schaffer, and Christopher Heinicke aligned with him to flesh out the
theory of attachment that is understood today.
Ainsworth, in particular, believed that the infant’s contribution to the attachment
process was more than biological and included his or her own internal appraisal of the
mother’s behaviors (Ainsworth, Blehar, Waters, & Wall, 1978). The “Strange Situation,” a
20-minute laboratory test developed by Ainsworth, was the first attempt to scientifically
capture the activation of attachment system behaviors between mother and child (Ainsworth
et al., 1978). One-year-old children were exposed to two brief separations from their
mothers; the responses of both mother and baby were recorded and became the basis for a
categorical system of attachment that is still in use today. Most of the children in this study
responded to their mothers’ absence with some distress but, at her return, were rather quickly
comforted and returned to their play. These babies were thought to be “securely attached.”
About 25 percent of the babies responded to mother’s return with indifference, a category
named “insecure—anxious avoidant.” Another 15 percent sought proximity to their mothers
but displayed little or no relief from their distress when reunited. This style was also
regarded as insecure, but called “anxious resistant.” The discovery that physical separation
alone could not account fully for infant response took attachment research to a new level.
5
Ainsworth and Bowlby persevered in their investigation of the idea that cognitive
mechanisms underpinned the behavioral components of the attachment system. Bowlby
coined the term “internal working model” to describe a process of mental representation that
the preverbal infant developed of his primary caregiver. In his historical work, Attachment
and Loss: Volume I, Attachment, Bowlby described a child’s “internal world” in this way:
Starting, we may suppose, towards the end of his first year, and probably especially actively during his second and third when he acquires the powerful and extraordinary gift of language, a child is busy constructing working models of how the physical world may be expected to behave, how his mother and other significant persons may be expected to behave, how he himself may be expected to behave, and how each interacts with all the others. Within the framework of these working models he evaluates his situation and makes his plans. And within the framework of the working models of his mother and himself he evaluates special aspects of his situation and makes his attachment plans (1969; pg. 354).
Using this concept, Ainsworth’s infants must have had distinct internal representations of
their mothers and of what separation from her meant. What went unmentioned in
Ainsworth’s original study was her ability to anticipate each infant’s attachment style based
on observations of maternal behavior. Her years of home observations in Uganda had caused
her to suspect a predictive link might exist between maternal responsiveness and security of
infant attachment (Ainsworth & Marvin, 1995). Ainsworth and colleagues subsequently
introduced the concept of “sensitivity” to describe the type of caregiving she found that
correlated with secure attachment on the part of the infant (Ainsworth, Bell, & Stayton,
1974). Mothers who exhibited sensitive caregiving behavior were those able to (a) attune to
infant’s signals with attentiveness, (b) appropriately interpret the signals, (c) respond
appropriately to the signals, and (d) react promptly, in a time period that did not provoke
excessive frustration for the child. Highlighting that central to the internal working model
was the expected availability and response of the attachment figure injected an importance
6
into the actions of the maternal part of the dyad that the previous systemic view of
attachment behaviors had not (Sroufe & Waters, 1977; Bowlby, 1973). Bowlby continued
to refine his approach, further hypothesizing that a child’s internal working model of self
takes a complementary position to the representation the infant has of his caregiver. In the
most primary form of this collaboration, the child evolves a representation of how acceptable
or unacceptable he is by how he feels his caregiver views him. More complex forms of this
transaction appear all through life in self-other relationships (Fonagy, 2001b).
Those who followed Bowlby and his fellow pioneers of theory moved beyond infancy
and began exploring attachment through the internal worlds of young children (Main,
suicidal acting-out, and vulnerability to psychopathology in childhood (Brisch, 2002). As
research on attachment disorders continues, new importance is ascribed to early identification
and intervention. Inge Bretherton aptly applies one of Freud’s statements:
So long as we trace the development from its final outcome backwards, the chain of events appears continuous, and we feel we have gained an insight which is completely satisfactory or even exhaustive. But if we proceed in the reverse way, if we start from the premises inferred from the analysis and try to follow these up to the final results, then we no longer get the impression of an inevitable sequence of events which could not have otherwise been determined (Bretherton, 1992; Freud, 1955).
It is an estimable goal to have enough knowledge about the role of MFA to “proceed in the
reverse way,” and endeavor to make the “inevitable sequence of events” in incomplete
mother-child attachment not so inevitable.
Poor attachment has not surprisingly been associated with the painful topic of fetal
and child abuse. A study in England with a sample of 40 women referred by Social Services
departments suggested that “negative preoccupied” antenatal attachment (as measured by the
Maternal Antenatal Attachment Scale) was predictive of an increased likelihood of
symptoms of anxiety, mood disturbance, and depression, self-reported irritation with the
fetus, and even fetal abuse (Pollock & Percy, 1999). Other researchers have looked at the
association between insecure attachment in mothers and the incidence of child abuse and
found positive correlations (Moncher, 1996). Contrastingly, strong MFA has been associated
with positive health practices during pregnancy, such as abstinence from tobacco, alcohol,
and illegal drugs, obtaining prenatal care, healthy diet and sleep habits, adequate exercise,
use of seat belts, and learning about pregnancy, childbirth, and infant care (Lindgren, 2001;
Lindgren, 2003).
21
Quality of attachment has also been associated with the perinatal mental health of the
mother. Weak attachment and negative maternal attitude have been associated with
postpartum anxiety (Blumberg, 1980; Gaffney, 1989) and depression (Condon et al., 1997;
Lindgren, 2001). On the other hand, strong attachment was found to be a moderator of the
vulnerability to postpartum depression in one sample of women in Israel (Priel & Besser,
1999). Personality vulnerability factors to depression were measured, and highly self-critical
women reported less depression when strongly attached to the fetus during pregnancy.
However, as reported in one integrative review, associations between attachment and
psychosocial variables have been disappointing (Cannella, 2005). Methods used across
studies have been inconsistent, psychometric properties of all instruments have not been
consistently valid and reliable, and the relationships investigated were exploratory rather than
theoretical. It was concluded that correlational studies utilizing theory-driven variables are
necessary for more significant findings.
An important factor for consideration is the large gap in existing research with
diverse populations. Psychometric data of current prenatal attachment measures has largely
been established using samples of low-risk, middle-class, American, Caucasian pregnant
women (Shieh et al., 2001). An increasing number of samples of women with high-risk
pregnancies (defined in this work as “fetal anomaly and/or the presence of a chronic disease
or pregnancy-induced disease threatening maternal or fetal health and carrying an increased
chance of mortality for either mother or fetus”) are being included in research, but only a few
published studies have included risk serious enough to require hospitalization (Penticuff,
1982). The reliability and validity of the existing tools for high-risk women is unknown, as
22
are the consequences of risk on parental adaptation and patterns of attachment. In addition to
the health/mortality concerns, high risk pregnancies include the significant possibility of fetal
anomaly and/or extreme prematurity. Extreme prematurity has been associated with a higher
incidence of insecure attachment; ongoing longitudinal work is examining this further, taking
into consideration the neurobiological risk factors (Brisch, 2002).
Finally, the children of societies everywhere deserve mothers (and fathers) prepared
for the awesome challenge of loving and training new humans. Five decades of research has
emphasized that caregiver response is the central element in how a child understands self and
others. John Bowlby expressed this most cogently:
Just as children are absolutely dependent on their parents for sustenance, so in all but the most primitive communities, are parents, especially their mothers, dependent on a greater society for economic provision. If a community values its children it must cherish their parents (Bowlby, 1951).
PURPOSE OF THE STUDY
In line with the recommendations of the works cited, this research will undertake a
theory-driven approach to the study of attachment in a sample of women hospitalized with
high-risk pregnancies. The purpose of this study is to investigate the influence of an
expectant mother’s personality style, her maternal object representations, and her ability to
develop an internal representation of her fetus upon the quality and intensity of maternal fetal
attachment. The exploration of the power of object representation and personality to predict
prenatal attachment in the context of hospitalization during high-risk pregnancy would have
relevance in both intervention and standard of care for such mothers. Since this is a highly
23
specialized sample of individuals, it is expected that other variables may have predictive
value. Prior depressive episodes, current levels of depression, type and severity of maternal-
fetal risk, as well as gestational age of the fetus at the onset of complications will be taken
into account.
The construct of MFA discussed in this work will be identified as suggested by Doan
and Zimmerman: “Prenatal attachment is an abstract concept, representing the affiliative
relationship between a parent and fetus, which is potentially present before pregnancy, is
related to cognitive and emotional abilities to conceptualize another human being, and
develops within an ecological system” (2003). Working on the assumption that MFA exists,
instruments developed from the psychodynamic approach will be employed to examine the
cognitive and emotional abilities of an expectant mother to conceptualize her own mother as
well as her fetus. Additionally, in view of the proposition that emotional factors preexisting
pregnancy are important potential determinants of prenatal attachment (Doan et al., 2003;
Mikulincer et al., 1999), the contribution of personality variables will be examined (Blatt,
Shahar, & Zuroff, 2001; Priel et al., 1999). The “ecological system” in this work will consist
of the situation of hospitalization due to high maternal or fetal risk, defined earlier as an
increased probability of fetal anomaly, compromises of maternal or fetal health, or maternal
or fetal demise. The literature review has been conducted consistent with these factors of
interest.
24
CHAPTER TWO
Review of the Literature
Since Cranley’s creation of a measure for her theoretical construct of maternal-fetal
attachment (MFA), there have been approximately 50 published studies incorporating some
measure of antenatal attachment in the research. The psychosocial variables examined have
included social support, interpersonal relationships, self-esteem/self-concept/sense of
mastery/efficacy, anxiety, depression, stress, and coping styles. Demographic characteristics
such as parity, age, level of education, and ethnicity have been incorporated into hypotheses
or analyzed post hoc. Biological variables such as previous substance abuse, maternal health
history, previous perinatal deaths, and maternal/fetal health outcomes have also been
correlated to attachment, and some studies have divided their sample by the presence or
absence of maternal-fetal risk. A few studies have concentrated on women with some level
of this risk, and even fewer have focused exclusively on women with risk severe enough to
require hospitalization. To date, there have been no examinations of the possible
relationships between personality style, object representations, severity of risk, and
attachment within a hospitalized population. However, there are bodies of research that
contribute significantly to our understanding of these factors in other contexts and with other
populations.
25
SEARCH METHODS
Relevant MFA studies published from 1981 (publication date of the Maternal Fetal
Attachment Scale) through 2005 were located through the use of various databases, including
Medline, Psychological Information, and Cumulative Index to Nursing and Allied Health
Literature. An ancestry approach (tracking research cited in studies reviewed) was also
employed to ensure the review of any articles missed in the database search. Dissertations
were excluded, as were articles written in languages other than English. The keywords used
were prenatal attachment, antenatal attachment, MFA, maternal-fetal attachment, internal
working model, object representation, mental representation, prenatal object relations,
personality, and high risk pregnancy. Abstracts of all articles supplied in the database
searches were reviewed to identify studies relevant to this review. Studies that incorporated
measures and discussions central to prenatal attachment, object representation, and
personality characteristics were all considered relevant. While this study is confined to
women who are hospitalized, studies that included non-hospitalized women were included in
light of the paucity of research with the hospitalized population. The studies reviewed are
categorized by their contribution to the understanding of the relationship of attachment to
object representation, personality, and risk.
INTERNAL WORKING MODELS AND OBJECT REPRESENTATION
Background
Bowlby’s premise of internal working models is reminiscent of the concept of object
representation proposed by Melanie Klein, from the very psychoanalytic approach that
26
Bowlby questioned. However, his concept of the “internal working model” has been
attributed to the psychologist and philosopher Kenneth Craik and his 1943 work, The Nature
of Explanation (Bretherton, 1992). Craik proposed that the human mind built “mental
models” of reality that were utilized to anticipate events and produce action. This internal
representational paradigm fit well with Bowlby’s belief that some cognitive mechanism was
at work with the biological system of attachment, for notice his connection in Attachment
and Loss Volume I: Attachment:
“If an individual is to draw up a plan to achieve a set-goal not only must he have some sort of working model of his environment, but he must have also some working knowledge of his own behavioural skills and potentialities… Henceforward the two working models each individual must have are referred to respectively as his environmental model and his organismic model… The environmental and organismic models described here as necessary parts of a sophisticated biological control system are, of course, none other than the ‘internal worlds’ of traditional psychoanalytic theory seen in a new perspective” (p.82).
The “new perspective” in this case had several facets. First, Bowlby held that these models
were based on real experience. Psychopathology resulted because a model “might become
totally out-of-date, or because it is only half revised and therefore remains half out-of-date, or
else because it is full of inconsistencies and confusions” but not because of unconscious
drives that generated fantasies that became internal representations (Bowlby, 1969). Second,
Bowlby pulled from the work of Spitz and Piaget to support his proposal that infants before
the age of nine months were not aware of the human characteristics of the “object;” in fact,
an infant could not even perceive of the “object” as having any permanence. He favored
Spitz’ term, “pre-object relation,” as well as Spitz’ idea that a smiling infant was responding
to a “visual gestalt signal,” not relating to a human. This fit in nicely with Bowlby’s thesis
that the five responses which made up attachment behavior—sucking, clinging, following,
crying, and smiling—were behavior patterns specific to man in much the same way that each 27
species in the animal world was endowed with its own peculiar repertoires of behavior.
Third, Bowlby also differentiated his term “instinctual response” from the usage of the term
“instinct” in psychoanalytic terms. Whereas psychoanalytic instinct referred to a
motivational drive, Bowlby’s term referred to an “observable pattern of behavior” that served
the evolutionary purpose of survival (Bowlby, 1986). Bowlby believed that during
maturation these early behavior patterns would move between various states of latency and
activity, being used in “fresh combinations.” Even the infantile behaviors like crying and
clinging would re-emerge in situations of danger, illness, or helplessness. Stress and
uncertainty could compromise adult-acquired defenses and infantile internal working models
would serve as defaults to guide behavior.
On the other hand, Melanie Klein’s conceptualization of “internal objects” (“object”
referring to a significant person in an emotional relationship) came from the psychoanalytic
view that Bowlby believed needed a “new perspective.” Klein extended Freud’s role of
fantasy as a specific mental process provoked by frustration to that of an elaborate collection
of unconscious images and knowledge that are the core of all mental processes. Klein
theorized that the child’s mental life is filled with fantasy as he establishes a complex set of
internalized object representations based on his experiences with primary caregivers. These
fantasies and anxieties concerning the internal objects become the underlying basis for an
individual’s behavior, emotions, and sense of self. Like Freud, she fueled these fantasies
with hypothesized libidinal drives and oedipal conflicts; unlike Freud, the drives are
Another contrast between attachment theory and object relations theory is how
research for each of these models was conducted. In the psychoanalytic world, investigations
of object representation were usually based on clinical case studies. Since quantitative
research was not an appropriate venue for exploring abstract conceptualizations of drive
theories, most writings were of clinical experiences with patients, often quite ill individuals.
Attachment theory, on the other hand, essentially came from work and study with normal
children and adults. The influence of ethological and biological research on these theorists
propelled them into empirical based research methods that had far greater generalizability
than the case-study driven work of the psychoanalysts.
Integration of Theory
For decades differences like these between attachment theorists and psychoanalytical
object-relational theorists prevented them from engaging in collaborative dialogue. In recent
years, there has been movement toward an integration of ideas made possible by several
changes. Peter Fonagy outlines these as: 1) A trend in attachment theory to move focus from
infant behavior and external determinants to greater interest in internal representations in
both infant and parent; 2) growing importance for observational and empirical research in the
psychoanalytic community, due to an awareness of the shortage of models that are both
scientifically acceptable and relevant to clinicians; 3) an openness between theoreticians to
integrated approaches and new ideas; and 4) the realization on the part of attachment
29
theorists that without growth through the integration of other approaches the attachment
model would remain stunted in terms of providing value to clinical work, enriching research,
and developing new theory (Fonagy, 2001a). Integrating Bowlby’s internal working model
with object representation is such an example of this new landscape.
Today’s understanding of internal working models is an amalgam of the contributions
of Bowlby, Ainsworth, Bretherton, Mulholland, Crittenden, Main, and Sroufe (Fonagy,
2001a). Fonagy depicts their composite description of the internal working model as an
evolution of four basic components: 1) Expectations based on transactions with and
interactive attributes of the primary caregiver created in the first year, 2) event
representations generated by general and specific memories of attachment-related
experiences, 3) autobiographical memories by which specific events are connected and
contribute to an ongoing personal narrative, and 4) the resulting inferential understanding of
the psychological characteristics of others and self (p. 14). Moving past the strict
interpretations of an internal working model as an imprint of historical relational
interchanges and an object representation as an elaborate concoction of libidinal-driven
fantasy allows a fresh paradigm: Mental representations and internal working models can be
described as reflections of reality-based early experiences with caregivers that are colored by
internal perceptions and transformations (Priel & Besser, 2001). Not ignoring the gap
between attachment and object relational theories, it has been proposed that these two
concepts overlap in a fashion that allows the quality of mothers’ mental representations to
predict the mother-infant attachment (Levine et al., 1991). This reconciliation of approaches
30
has inspired new forays in research with clinical relevance, particularly in the discussion of
MFA.
MFA and Object Representation
Attachment
Bowlby himself believed that the transition of a woman to motherhood mobilized the
same “forces” that had in early infancy and childhood attached her to her own mother (1986).
Some years later, Rubin restated this in her early discussion of the tasks required to fully
attain the maternal role, reporting “Mother was a major prototype and was the most
significant contributor of subject’s set of anticipations in becoming a mother” (Rubin, 1967).
If the participant’s mother was deceased, she was either interjected into the interviews in the
form of memory, or augmented by an aunt, mother-in-law, grandmother, or a person of the
mother’s generation who may have had a maternal relationship with the subject. Rubin
likened this to the “binding-in” task with the fetus, stating that her subjects seemed to also be
“binding-in” again with the mother or mother substitute. In cases where the expectant
mother was separated from her mother by distance, Rubin found that often there was an
actual or “wished-for” trip that seemed almost like a “pilgrimage.”
While not specifically looking at MFA, the Fonagy, Steele, and Steele (1991)
research appears to be the first that empirically explored the association between adult and
infant attachment style in a prospective manner beginning in pregnancy. In this longitudinal
study, the Adult Attachment Interview (AAI) was administered to 100 primagravidas in their
last trimester of pregnancy. At 12 months postpartum, the mothers and infants were assessed
in the Strange Situation exercise. An impressive 75% of mothers categorized as secure had
31
securely attached children; 73% of mothers classified in one of the insecure descriptions had
insecurely attached children. A second finding relevant to the discussion of object
representation in internal working models is that the quality of those relationships could be
measured by the expectant mother’s ability to articulate a complex representation of the
expectant mother’s relationships with her parents. These women were able to:
“fluently convey a global representation (whether favorable or unfavorable) of what her relationship to each parent was like during her childhood…she demonstrates an understanding of her own personal development that includes an awareness of the multiple motives (conscious and perhaps unconscious) that guided her parents’ behavior toward her…there are no significantly distorting mental processes at work (pg. 901).
These robust findings were not totally unexpected; Mary Main, one of the developers
of the AAI, had herself wondered if adult attachment interviews might have something to say
about the mechanism behind the intergenerational cycle of child abuse (Main & Goldwyn,
1984). In a study of 30 normal, non-abusive women whose children had 4 years earlier
participated in an Ainsworth Strange Situation study, Main found that a mother’s experience
of her own mother as rejecting was related to her rejection of her own infant. In addition,
these women also revealed systematic cognitive distortions, such as idealization of the
rejecting parent, difficulty in remembering childhood, and incoherency in discussing their
attachment to their mother. Main found one exception: Women who could coherently
describe their rejection by their mothers, expressing resentment and anger, did not exhibit the
same avoidant behavior toward their own infants. Evidently, the differences in attachment
patterns in these children and women were also related to important cognitive differences, as
illustrated by the degree of cohesion and consistency the adults were able to utilize in their
narratives (Main et al., 1985).
32
Integrating Main’s findings, Fonagy’s team hypothesized that internal working
models become activated by certain expectations or events, influencing attachment-related
cognitions and behaviors that may be best thought of as “attachment states” (Fonagy et al.,
1991). These are distinct from the internal working model which, along with personality
traits, predisposes individuals to feelings and behaviors. Their recommendation was that
models of attachment could be informed by the examination of representational processes
that influence attachment-related emotions, thoughts, and behaviors.
The AAI and the Strange Situation have been paired in research repeatedly since the
Main and Fonagy studies, with findings suggesting that attachment classifications are stable
even across three generations (Zeanah et al., 1993; Benoit & Parker, 1994). Meanwhile,
interest in MFA began increasing in the field of study devoted to child and adolescent mental
health. A pair of child psychiatrists introduced the Working Model of the Child Interview
(WMCI), a measure designed to evoke mothers’ internal representations of their infants.
Modeled after the AAI, this one-hour structured interview categorically scored a caregiver’s
perceptions and subjective experience of their child. Qualitative, content, and affective
features of the narrative result in the assignment of one of three classifications: Balanced,
disengaged, or distorted. “Balanced” responses convey relatively rich details about the
infant, including both positive and negative characteristics of the infant or the mother-infant
relationship. “Disengaged” narratives are cool, distant, or indifferent descriptions that
implied the infant’s experience was either unrecognized or disregarded; descriptions are
unelaborated, giving the sense that the caregiver does not truly know the child. “Distorted”
representations are internally inconsistent, confusing, unrealistic, and divulge a lack of
33
insight concerning the impact of parenting upon the infant. In two independent
investigations with samples of mothers and their 12-month-old infants, they found that the
mothers’ WMCI classifications were significantly correlated to their infants’ attachment
classifications according to the Strange Situation. Narratives classified as “balanced” were
associated with secure infants, “disengaged” with insecure-avoidant babies, and “distorted”
with insecure-resistant children. Benoit et al. replicated the study with 96 expectant mothers
in the third trimester of pregnancy (Benoit, Parker, & Zeanah, 1997). When the WMCI was
repeated one year later concurrent with the Strange Situation, not only were WMCI
classifications significantly stable over time for the mothers, but the pregnancy WMCI
results predicted infant Strange Situation classifications in 74% of the cases. Concordance
between 12-month WMCI and Strange Situation classification was 73%. The authors felt
that this association could have major implications for early identification of high-risk
parent-infant relationships.
Only two other research teams to date have published empirical studies of the
relationship between MFA and object representations. Levine et al., working with a sample
of 42 pregnant adolescents, hypothesized that object representation during pregnancy could
be a predictor of infant attachment style (1991). The AAI was administered before childbirth
and the Strange Situation was conducted when the 42 infants were 15-months-old. The
quality of object representation was measured by the application of the Krohn Object
Representation Scale for Dreams to the responses of the AAI. The Krohn measure is an 8-
point scale that assesses hierarchical levels of an individual’s maturity of object-
representation and capacity for interpersonal relatedness (Krohn & Mayman, 1974).
34
Originally developed for use with dream analysis, it has been used in a similar way with the
responses of a Rorschach administration; however, there is no information as to its reliability
and validity in the application of it to the AAI. While the method has been questioned (Priel
et al., 2001), this research team did find mothers’ attachment style and object representations
highly correlated, and also found that both were significantly correlated with infant’s
attachment style as categorized at 15 months of age. However, the extent to which these
findings are generalizable is questionable in light of the fifteen-year mean age of this
population. Even considering individual variation in the development of mental
representations, this process is usually viewed as unfolding gradually through maturation to
adulthood and subject to disruption by developmental demands that are age-inappropriate
(Blatt, 1974).
Overlapping Models
What exactly has been measured in this body of research? Since the AAI has formed
the basis for these analyses, is the discussion so far one of internal working models or of
object representation? While the two concepts are related, they are not identical, and as such,
may not have the same contribution to MFA. Internal working models might be described as
templates of relationships (Main et al., 1985). In contrast, object representations reflect the
internal transformations of early relationship patterns, thereby guiding the use of and
affecting the organization of those templates (Fonagy et al., 1991). The AAI was designed
to elucidate early patterns of reality-based attachment relationships and the participant is
asked to reflect on what those mean to him or her currently. The WCMI was constructed on
the same theoretical base; their correspondence is perhaps not surprising. The findings of
35
these analyses may certainly be revealing what sort of template an individual is describing,
but an understanding of what links internal working models with antenatal or maternal-infant
attachment continues to be elusive for these researchers. Some theorists, following Main’s
concept of maternal sensitivity, have assumed that maternal responsivity/sensitivity mediates
maternal attachment and infant security (van Ijzendoorn, 1995). This assumption has not
been proven, however, and other attachment theorists point to analyses of discordant mother-
infant dyads and counter that sensitivity is more accurately thought of as a moderator that can
block an attachment state of mind, as when a mother with an insecure attachment style is
found to have a secure infant (Atkinson et al., 2005). Therefore, if the interest is in what
activates a template or internal working model perhaps another form of measurement is
needed.
With this in mind, Priel and Besser of Ben-Gurion University in Israel formulated and
tested their hypothesis that a pregnant woman’s antenatal attachment and internal working
models would be mediated by the expectant mother’s internal representation of her own
mother on a sample of 120 first-time mothers in the third trimester of a low-risk pregnancy
(2001). Their strategy was to utilize operationalizations of internal working models and
object representations that corresponded to Bartholomew & Horowitz’ conceptualization of
adult attachment and Blatt and colleagues’ means of identifying object representations,
instead of the traditional Ainsworth/Main constructs of adult attachment and internal working
models (Bartholomew et al., 1991; Blatt, Chevron, Quinlan, Schaffer, & Wein, 1992a). The
Bartholomew and Horowitz conceptual development was an outgrowth of Hazan and
Shaver’s application of George and Main’s attachment categories of the AAI to the romantic
36
attachment that develops between adults (Hazan et al., 1987; George et al., 1985). As a
result of noticing a difference in the motivations behind avoidance when exhibited by
dismissive behavior (George and Main’s category) as opposed to fearful behavior (Hazan and
Shaver’s category), Bartholomew & Horowitz hypothesized a classification of adult
attachment that is defined by the positivity of the internal working models of self and other.
Four dimensions of positivity/negativity of self/other could explain four models of
Unlike most previous studies utilizing largely Caucasian, middle-class, married
women, this project enjoyed a richly diverse sample. Table 1 provides the demographic
information for the sample. The sample consisted of 91 women, ranging in age from 17 to 44
years, with a mean age of 27.01 (SD = 6.44). Fifty-four percent (N = 49) were Caucasian,
33% were African American (N = 30), 11% were Hispanic (N = 10), and 2% were Asian (N
= 2). Forty-nine percent of the participants were married (N = 49), 32% were single (N =
29), 10% were living with a partner (N = 9), and 3% were separated from their husbands (N =
3). The average number of children at home was 1.02, with a range of 0 to 6 children (SD =
1.45). Eleven per cent of the sample did not complete high school (N = 10), however 25%
completed high school or had completed a General Educational Development test (N = 23),
33% received some college education (N = 30), and 26% had obtained an undergraduate or
graduate college degree (N = 24). Thirty percent had an annual household income below
$26, 000 (N = 27), 16.5% reported income of $26,000--$40,000 annually (N = 15), 16.5%
reported $41,000--$65,000 (N = 15), and 29% reported earnings of more than $65,000 a year
(N = 26). Government funds (Medicaid) covered 43% (N = 15), 48% were covered by a
private insurance plan (N = 44), and 3.3% had no coverage at the time of their admission to
Baylor (N = 3).
66
Pregnancy Characteristics
Although almost one-third of the sample was experiencing a first pregnancy (N = 29),
24% had been pregnant before (N = 22), and 44 % (N = 39) had history of at least two
previous pregnancies (Table 2). Four percent (N = 4) had suffered delivering a stillborn
baby, 27% had history of miscarriage (N = 24), and 37% had history of obstetric
complications (N = 34). Almost half of the patients were in their third trimester at onset of
obstetric complications (N = 42), but 11% (N = 10) were in their first trimester and 37% in
their second (N = 34).
Psychiatric Characteristics
As Table 3 reveals, slightly more than three-quarters of the participants denied any
history of psychiatric disturbance, hospitalization, psychotropic treatment, or counseling (N =
69), but 7% reported a history of depression (N = 6), 7% reported a history of anxiety
disorder (N = 6), 3% had been treated for both depression and anxiety (N = 3), and one
participant suffered from bipolar disorder. However, when screened for depression, 36% (N
= 33), endorsed depressive symptoms on the Center for Epidemiologic Studies- Depression
scale (CES-D; score exceeded the threshold of 16 for possible mood disorder) and 42% (N =
38), endorsed depressive symptoms on the Edinburgh Postpartum Depression Scale (EPDS).
Although 35 Structured Clinical Interviews were administered, only three patients met the
criteria for Major Depressive Disorder (six met criteria for one of the anxiety disorders).
Seventy-seven percent (N = 70), of the participants denied a familial history of psychiatric
disturbance, but 18% (N = 16), reported family history of mood, anxiety, or substance-use
disorders.
67
Nature of Obstetric Risk
There was significant variance in the severity of risk across participants in the sample
(Table 4). The mean revised Hobel Assessment score was 18.35 (SD = 9.62), with a range of
5 to 45 points. Twenty-five percent of the patients fell between 10 and 15 points of severity.
Classifying risk as “maternal” or “fetal” was not problematic; however, 42% (N = 38) of the
sample met criteria for both types of risk. This would occur, for example, when a participant
would carry the diagnosis of toxemia (severe gestational hypertension) and would also have a
history of a stillbirth and/or a premature delivery. In nine of these instances (21%), the dual
risk was due to a multiple pregnancy.
OVERVIEW OF STATISTICAL ANALYSES
Primary Hypotheses
Hypothesis One
Participants who expressed an integrated understanding of their mother’s
characteristics, qualities, and aspirations in the Object Relations Inventory (ORI) narrative
were expected to also report a higher quality and greater intensity of attachment, as evaluated
by two factors of the Maternal Antenatal Attachment Scale (MAAS). The ORI narratives are
scored on six dimensions: Benevolence, Punitiveness, Ambition, Length, Ambivalence, and
Conceptual Level. The subject’s responses on the MAAS yield a global score of attachment
as well as scores on two orthogonal factors: Quality, describing positive affect regarding the
fetus, and Intensity, reflecting the amount of time the expectant mother reports being
preoccupied with thoughts about the fetus. A one-way multivariate analysis of variance
68
(MANOVA) was conducted to determine the effect of the six dimensions of the ORI upon
the MAAS global score of attachment. The scores were separated into two levels, above and
below the mean (M = 81.5, Table 4), in the first analysis. No significant differences were
found among the dimensions for attachment, Wilks’ Λ = .899, F (6, 32) = .601, p = .728.
The multivariate η² based on Wilks’ Λ was nonsignificant, .101. Table 5 contains the means
and the standard deviations on the dependent variables for the six groups. A second
MANOVA was conducted, dividing the Global score into three groups: Lowest through -.99
sd below the mean, -1 sd through 1 sd, and 1.01 sd through the highest score. This further
stratification was also nonsignificant, Wilks’ Λ = .620, F (12, 62) = 1.393, p = .193, η² = .212
(Table 6).
A one-way analysis of variance was conducted to evaluate the relationship between
Conceptual Level of the mother narrative and the Quality of attachment score from the
MAAS. The 95% confidence intervals for the multiple comparisons, as well as the means
and standard deviations for the four Conceptual Levels, are reported in Table 7. The
independent variable, Conceptual Level of the narrative, included four levels: “Sensorimotor-
Preoperational,” “Concrete,” “External Iconic,” and “Internal Iconic” (no narratives
contained the necessary elements for the highest level, “Conceptual Level”). The dependent
variable was the Quality of Attachment score from the MAAS. The ANOVA was
nonsignificant, F (3, 79) = .434, p = .729. Figure 1 displays the distribution of the sample.
The analysis was repeated using the attachment Intensity score from the MAAS as the
dependent variable, with similar results: F (3, 77) = 1.35, p = .265 (Table 8; Figure 2). A
significant issue in the interpretation of the preceding analyses is the uneven distribution of
69
the sample across the Conceptual Level dimension of the ORI (Table 9). In view of this
limitation, several other analytical strategies were employed.
Correlation coefficients were computed among the Quality of Attachment factor, the
Intensity of Attachment factor, and only the Conceptual Level of the mother narrative (Table
10). Since the Conceptual Level score is ordinal, Spearman’s rho was computed for this
comparison. The results of the correlational analysis were nonsignificant (.02 for Quality
and -.210 for Intensity). See Figures 3 through 5 for distributions of Conceptual Level across
the MAAS Quality, Intensity, and Global scores.
In the following analysis, the ORI scores for Conceptual Level were collapsed into
two groups: Participants with narratives in the Sensorimotor-Preoperational and Concrete
levels were combined and those with narratives in the External and Internal Iconic level were
combined. A two-way contingency table analysis was conducted to assess whether there
were differences in the Quality of Attachment factor between the two levels of narratives
(Table 11). The Quality scores were divided above and below the mean and, once more, test
results were nonsignificant: Pearson χ² (1, N = 83) = .855, p = .465. A second analysis was
undertaken, forming three groups of attachment scores: “Low,” lowest through -.6,
“Average,” -.5 through .5, and “High,” .6 through the highest score (Table 12). (Since there
were no scores greater than +1 standard deviation from the mean it was necessary to choose
.5 as the point for analysis.) The test statistic changed slightly: Pearson χ² (2, N = 83) = .111,
p = .946 and, once again, the problem of unbalanced distribution across the groups of
attachment calls for caution in interpretation.
70
These two analyses were repeated exploring the relationship between MAAS
Intensity of Attachment scores and Conceptual Level of the mother narrative. The first two-
way contingency table analyzed two levels of object representation Conceptual Level,
Preoperational/Concrete and External/Internal Iconic, and two levels of Intensity, above and
below the mean. Table 13 displays the nonsignificant results: Pearson χ² (1, N = 81) = .173,
p = .727. The second analysis repeated the stratification of Conceptual Level and also
stratified Intensity into three levels: “Low” (lowest through -.6), “Average” (-.5 through .5),
and “High” (.6 through the highest score). These nonsignificant findings appear in Table 14:
Pearson χ² (1, N = 81) = 2.04, p = .361. In light of these findings, no further analyses were
performed. The results suggest that the participants’ mental representations of their mothers
were not related to the quality or intensity of antenatal attachment as evaluated by the
MAAS.
Hypothesis Two
Originally, it was also proposed that these same analytic procedures would be carried
out with participants’ narratives about their fetuses. However, in scoring the narratives it
became apparent that the twelve characteristics from which the Benevolent, Punitive, and
Ambitious dimensions are derived were not appropriate for a narrative about an unborn baby.
Three dimensions, Ambivalence, Length, and Conceptual Level, were retained for analysis,
and a MANOVA was conducted with these dimensions and the MAAS Global Attachment
score (once again divided into an “Above the Mean” group and a “Below the Mean” group).
A significant difference between the groups was found, Wilks’ Λ = .843, F (3, 81) = 5.031, p
= .003, η² = .157. Means and standard deviations are contained in Table 15. Analyses of
71
variances (ANOVA) on each dimension were conducted as follow-up test to the MANOVA.
The ANOVA on the Ambivalence dimension was significant, F (1, 83) = 11.12, p = .001, η²
= .118, while the ANOVAs on the Length, F (1,83) = .026, p = .873, and Conceptual Level,
F (1, 83) = .003, p = .958, were nonsignificant. When subjected to the same second analyses
with the Global Attachment score categorized in three levels, the significance remained;
Wilks’ Λ = .769, F (6, 160) = 3.735, p = .002, η² = .123 (Table 16). In the follow-up
ANOVAs on each dimension, the statistic on Ambivalence increased in significance, F (2,
82) = 8.52, p = .000, η² = .172. Baby narrative Conceptual Level and Length remained
nonsignificant. These findings suggest that mothers who express ambivalent feelings about
their babies have slightly lower Global scores of antenatal attachment.
It was hypothesized that women who could articulate more complex representations
of their mothers would also be able to do so of their babies, as evaluated by the Conceptual
Level dimension of the ORI. Taking into consideration the ordinal quality of the measures, a
Spearman’s rho analysis was performed and no significant association between the
Conceptual Levels of the two narratives was suggested, r = .068, p = .55. Figure 6 illustrates
the distributions of Conceptual Levels of the two narratives. Two other dimensions of the
two narratives, Ambivalence and Length, were also compared. The results of the
correlational analyses presented in Table 17 show that 8 out of the 15 correlations were
statistically significant, ranging from r = .27 to r = .72. All significant correlations were
related to Ambivalence or Length, one of which was associated with Conceptual Level. The
findings seem to suggest that women who express ambivalence toward their mothers also
express it of their babies, and these narratives tend to be more fluent than those in which no
72
ambivalence is scored. The length of the baby narrative was positively associated with the
Conceptual Level of the baby narrative, perhaps not surprising in that higher Conceptual
Levels would require greater articulation.
Two bi-level variables were created by collapsing the Conceptual Levels of mother
and baby narratives into two categories, Preoperational/Concrete and External/Internal
Iconic. A two-way contingency table analysis was conducted to evaluate whether there were
associations between these broader categories of representation. The results of the test were
nonsignificant, Pearson χ² (1, N = 82) = .206, p = .695 (Table 18). Although the two
narratives may resemble one another in terms of Ambivalence and Length, the findings do
not support any association on the dimension of Conceptual Level.
Hypothesis Three
According to Condon’s model of antenatal attachment, expectant parents resided in
one of four quadrants of attachment style, estimated by the scores on the Quality and
Intensity factors (Condon, 1993). The final stated expectations of the primary hypotheses
were that dependent tendencies would be more highly associated with an anxious, ambivalent
style of prenatal attachment as conceptualized in the fourth quadrant of the MAAS
Attachment Style paradigm. Self-critical tendencies, as identified by scores above the mean
on the Depressive Experiences Questionnaire (DEQ) were expected to be more highly
associated with a high quality of attachment but an avoidant style (second quadrant).
Blatt’s original scoring method of the DEQ (raw scores are transformed into z scores,
weighted according to the factor they are most highly correlated, and summed) was
employed for these analyses (Blatt et al., 1976). Means and standard deviations of the DEQ
73
are located in Table 4. First, a Pearson’s product-moment correlation was performed
between the DEQ z scores for dependency and self-criticism and the MAAS Global, Quality,
and Intensity of Attachment scores (Table 19). The self-critical scale of the DEQ was
negatively correlated with the MAAS Quality of Attachment factor (N = 91; r = -.366, p <
.000), but analysis revealed no significant association between the self-critical scale and the
MAAS Intensity of Attachment factor. Dependency was not found to correlate with either
the quality or intensity of attachment.
In the next analyses, participants were identified by one of Condon’s quadrants
according to their scores on the Quality and Intensity factors of the MAAS. Following
Condon’s theoretical framework, four categories were established: Quadrant One, high
Quality and high Intensity (strong, secure) ; Quadrant Two, high Quality and low Intensity
(avoidant); Quadrant Three, low Quality and low Intensity (withdrawn); and Quadrant Four,
low Quality and High Intensity (anxious, ambivalent). Quality and Intensity were rated
“high” when above the mean of the sample and “low” when below the mean. Figure 7
portrays the distribution of the sample according to Condon’s formulation. In order to
identify highly dependent or highly self-critical individuals from the DEQ scores, it had been
proposed that the sample be divided into individuals within and outside of two standard
deviations of the mean. After the data were collected, it became apparent that this strategy
was not the best way to analyze the data because of the homogeneity of the scores.
Therefore, differences of greater or less than one standard deviation on the DEQ scores
established three categories of the Self-critical style (M = -.93, SD = 1.01) and Dependent
style (M = -.53, SD = .90). (Categories were “Low” < -1 standard deviation from the mean,
74
“Average” > -1 but < 1 standard deviation from the mean, and “High” > 1 standard deviation
from the mean. A contingency table analysis was then conducted to investigate whether
pregnant women who are more self-critical reported a higher quality of attachment but a
lower intensity (Quadrant Two). A Pearson χ² analysis was nonsignificant, (6, N = 91) =
5.714, p = .456. However, the sample size was not large enough to afford a valid
interpretation (Table 20). Therefore, the Self-Criticism factor was condensed to two levels,
below and above the mean (M = -.93). The sample still lacked power, however the Pearson
χ² was significant, (3, N = 91) = 8.93, p = .03, η² = .21 (Table 21). Nevertheless, a look at the
sample reveals that those individuals who are more self-critical are less likely to reside in the
Avoidant quadrant than in the other three. It is unlikely, even with greater power, the
hypothesis would be supported.
The same approach was attempted with Dependency scores. In this sample, so few
participants scored in the average range of DEQ Dependency that the resulting Pearson chi-
square analysis cannot be interpreted (Table 22). The second analysis, identifying subjects as
above or below the mean on the Dependency factor, improved the distribution of the sample,
but those who endorse dependent statements do not appear to have a clear preference for any
of Condon’s four quadrants (Table 23). Although the analyses are somewhat underpowered,
these findings suggest that women who are more self-critical report a lower quality of
attachment and trend toward the third (withdrawn) and fourth (anxious ambivalent) quadrant
attachment styles in Condon’s model. There appear to be no significant differences in the
level of attachment intensity such women report. On the other hand, the attachment scores of
women who are more dependent do not to fit clearly into any of the Condon quadrants.
75
Hypothesis Four
It was further expected that women with a self-critical personality style would convey
object representations of a higher Conceptual Level and report a higher quality of attachment
than women with a dependent style. Once again, the distribution of the sample across the
categories did not allow for a valid analysis of four Conceptual Levels of the narrative by
self-criticism or dependency in three categories, Low, Average, and High (Tables 24 and 25).
Even when stratifying the narrative Conceptual Level into two levels
(Preoperational/Concrete and External/Internal Iconic) and the DEQ Dependency and Self-
Critical factors above and below the mean (Tables 26 and 27; Figures 8 and 9), the sample
does not distribute as expected, and is not significant. For the Self-Critical model, the
Pearson χ² (1, N = 83) = 1.51, p = .272; for the Dependent model, the Pearson χ² (1, N = 83)
= 1.32, p .284. However, the data trend toward a refutation of the original hypothesis.
Secondary Hypotheses
Hypothesis Five
The secondary hypotheses are to be viewed as exploratory. Firstly, it was
hypothesized that fetal representation and attachment might be influenced by the type of risk
(maternal, fetal, or combined) described to the patient by her obstetrician. A one-way
analysis of covariance (ANCOVA) was conducted. The independent variable, Conceptual
Level of the ORI, included two levels: Preoperational/Concrete and External/Internal Iconic.
The dependent variable was the MAAS Global attachment score and the covariate was risk.
The ANCOVA was non significant, F (1, 82) = .05, MSE = 2.49, p = .823. Holding risk
76
constant, there was no relationship between Conceptual Level of the baby narrative and the
Global attachment score.
Secondly, it was expected that those mothers identified as being hospitalized for
significant risk of fetal demise would report higher antenatal attachment, as indicated by the
global score of the MAAS, than mothers hospitalized because of significant maternal risk
factors. The risk factors of the revised Hobel scale as identified primarily “fetal” or
“maternal” are listed in Tables 28 and 29. A one-way analysis of variance was conducted to
evaluate the relationship between type of risk and attachment. The independent variable,
type of risk, included three levels: fetal, maternal, and both fetal and maternal. The
dependent variable was the change in the Global score of the MAAS. The ANOVA was
nonsignificant, F (2, 90) = .436, p = .648. The means and standard deviations for the three
types of risk are reported in Table 30 and portrayed in Figure 11. Since the third group
incorporates fetal risk, the three groups were combined into two groups, maternal and
fetal/combined maternal-fetal. Another one-way analysis of variance was conducted and was
also nonsignificant, F (1, 90) = .04, p = .841. The means and standard deviations for the two
groups are reported in Table 31. These results suggest that the type of obstetric risk has little
association with mental representation or reported attachment.
Hypothesis Six
It was expected that mothers with identified fetal risk would report a greater intensity
of antenatal attachment. First, a one-way analysis of variance was performed with the
MAAS Intensity score as the dependent variable and type of risk the independent variable
(means, standard deviations, and pairwise comparisons appear in Table 32). The ANOVA
77
was nonsignificant, F (2, 86) = 1.239, p = .295. A two-way contingency table analysis was
then conducted with three levels of risk, maternal, fetal, and fetal-maternal, and two levels of
Intensity, above and below the mean (N = 89, M = 30.71, SD = 4.88). Fetal risk and intensity
of attachment were nonsignificantly related, Pearson χ² (2, 89) = 1.36, p = .506 (Table 33).
The MAAS Quality factor and Global score were also analyzed in this manner, with no
significant findings (Table 33). The analyses were repeated collapsing risk into two
categories, maternal and fetal/combined maternal-fetal. The results remained nonsignificant,
Pearson χ² (1, 89) = .048, p = 1.00 (Table 34). A Pearson’s product-moment correlation was
performed for risk severity (represented as the sum of Hobel weighted risk items) and the
three MAAS scores (Quality, Intensity, and Global). Table 35 reports the values. A one-way
analysis of variance was also executed placing the MAAS Intensity score as the dependent
variable and the level of risk as the independent variable (means, standard deviations, and
pairwise comparisons appear in Table 36). The statistic, F (2, 86) = 1.25, p = .293, was
nonsignificant. From these analyses, there does not appear to be a significant relationship
between type of risk or level of risk severity and intensity of attachment as measured by the
MAAS.
Hypothesis Seven
It was hypothesized that older gestational age of the fetus (calculated in weeks) at the
time of admission to the antenatal unit would be positively correlated with the Global
attachment score of the MAAS. A Pearson’s product-moment correlation was employed to
assess this relationship and was also nonsignificant, so this hypothesis is disproved. Table 37
reports the correlations and probabilities.
78
Depression and Attachment
As reviewed earlier, depression and antenatal attachment have been linked by
numerous previously published studies. The data from this population support those
findings. Both screening measures, the Edinburgh Postpartum Depression Scale (EPDS) and
the Center for Epidemiologic Studies-Depression scale (CES-D) were negatively correlated
with the Quality of Attachment factor from the MAAS (Table 36). The screening measures
do not have a significant correlation with the Intensity factor, and only the EPDS is
significantly correlated with the Global score.
In order to examine this association more closely, a univariate analysis of variance
was conducted with the EPDS score as the dependent variable and attachment style as the
independent variable. The F ratio was significant, 8.693 (3, 87), p = .000, and the
relationship strong, as assessed by η², with the Quality of Attachment score accounting for
23% of the variance of the dependent variable. Post-hoc tests were performed to evaluate
pairwise differences among the means (Table 40). There was a significant difference in the
means between the groups that reside in the High Quality quadrants of Condon’s attachment
style framework. Those participants who were above the mean on the Quality factor reported
fewer depressive symptoms than those who were below the mean and in the two Low Quality
quadrants of the model. A second analysis confirmed the findings, Pearson χ² (3, 91) =
21.339, p = .000. This relationship was stronger than that reported in the ANOVA (η² = .39).
79
Exploratory Analyses
DEQ: As some have been concerned with the use of factor-derived scale scores in the
original scoring system of the DEQ, several analyses were run on the variables of interest
utilizing other scoring procedures of the instrument (Bagby et al., 1994; Rude & Burham,
1995; Santor, Zuroff, & Fielding, 1997a; Blatt et al., 1995). Bagby et al. proposed a
shortened scale in which items with excellent fit to the two-factor model (Dependency and
Self-Criticism) were isolated (1994). Blatt and colleagues identified two “facets” of the
Dependency factor that suggested two different levels of interpersonal functioning:
“Dependence,” derived from items that endorsed feelings of helplessness, broad
apprehensions about rejection or fears of separation and loss not related to a specific
individual, and “Relatedness,” the product of items that consider feelings about the loss or
loneliness that might occur as the result of disruption in a relationship with a particular
significant person (1995). Rude and Burnham theorized that dependency was not entirely
pathological, and that the Dependency scale of the DEQ could be divided into the subscales
“Connectedness” and “Neediness” (1995). Connectedness referred to healthy valuing of
relationships and neediness applied to the pathological anxiety concerning rejection and loss.
Santor et al. introduced a shortened instrument that utilizes the preferred unit-weighted
composite scoring system yet preserves the orthogonality of the factors found in the original
scoring system (1997).
First, Pearson’s product-moment intercorrelations were computed for the sample
across all four scoring systems. Table 42 reports the means and standard deviations for the
sample and Table 43 reports the intercorrelations. As would be expected, the scoring
80
systems are highly correlated; out of 28 correlations, 14 exceed r =.60, and only 5 are
nonsignificant. Note both the original scoring and the McGill scoring of the Self-Criticism
factor are not associated significantly with Rude & Burnham’s “Connectedness” aspect of
Dependency but are moderately associated with the “Neediness,” the less healthy aspect of
interpersonal functioning.
Second, Pearson product-moment correlations between MAAS attachment scores and
the three revised scoring system were conducted (Tables 44-46). The Self-Critical factor, as
scored in the McGill system, is negatively related with Quality (r = -.339, p < .01). Blatt’s
revised subscale of Dependency, Relatedness, is negatively associated with Quality (r = -
.241, p < .05) and Rude & Burnham’s subscale, Neediness, is also negatively associated with
Quality (r =-.242, p < .05).
Finally, a series of regression analyses were then conducted to compare to what
extent each scoring method of dependency and self-criticism scores could predict the MAAS
Global Attachment score. The predictors were the eight scores (original, McGill, Blatt
revised, and Rude and Burnham revised), and the criterion variable was the overall measure
of attachment provided by the MAAS. Out of four analyses, no linear combination of
dependency and self-criticism scores proved to be significantly related to the attachment
index. Table 47 contains reports the results.
Gestational age: Other investigators, as cited in the literature review, have found
relationships between attachment and gestational age, particularly after quickening. No such
relationship was found in this population in a Pearson product-moment correlation of
81
gestational age (weeks) and MAAS Global Attachment score (r = .135, p = .203), Quality of
Attachment (r = .08, p =.45), or Intensity of Attachment (r = .157, p .142).
Object Relations Inventory: The use of the Object Relations Inventory in assessing
the maternal-fetal relationship is as yet unpublished. To explore a possible relationship
between the baby narrative and gestational age, a Spearman’s rho correlation was conducted
and no significant associations were found (Table 48). Multivariate analyses of variance
were also conducted on the Length, Ambivalence, and Conceptual Level dimensions of the
mother and baby narratives across ethnicity, with nonsignificant results: Wilks’ Λ = .77, F
(18, 204) = 1.09, p = .368. Pearson’s product-moment correlations were performed between
age and the above dimensions. All correlations were nonsignificant with the exception of
ambivalence in the mother narrative, r = .230, p <.05. Within this sample, younger women
revealed more ambivalent feelings in the narratives about their mothers than did older
participants.
Another analysis explored the possibility of a relationship between the baby narrative
and having other children. A Pearson’s product-moment correlation was conducted between
“number of children at home” and the Conceptual Level of the ORI. There was no
significance: r = -.054, p = .627.
Maternal Antenatal Attachment Scale: Table 39 reports an interesting positive
correlation between age and Quality of attachment (r = .29, p < .01). This finding suggests
that older women report a higher quality of attachment than do their younger colleagues. A
second exploration investigated parity, however a two-way contingency table found no
82
significant difference in Global attachment scores of multiparous women and primagravidas,
Pearson χ² (1, 88) = 2.69, p = .10.
83
CHAPTER FIVE
Conclusions and Recommendations
The purpose of the present investigation was to delineate the convergence of
personality, object representation, and antenatal attachment in the context of significant
maternal or fetal risk. Previous research in general populations has associated depression
with the personality tendencies of dependency and self-criticism, and also with lower levels
of development in object representation. As these relationships have been investigated in
women during the perinatal period, findings have suggested that not only are there
relationships between personality, object representation, and depression, but depression often
impinges upon the natural process of maternal attachment to baby (Priel et al., 2001).
Depression and attachment seem to have a reciprocal relationship, for other research has
suggested that strong antenatal attachment acts as a protective factor against postpartum
depression (Priel et al., 1999). Most prior research has focused on uncomplicated
pregnancies in a Caucasian, middle-class, married population, however studies that have
included women with elevated obstetric risk have suggested that such risk may be both a
moderate predictor of postpartum depression and a risk factor for healthy antenatal
attachment. This study attempted to assess the variables of personality, object representation,
and maternal/fetal risk with the expectation they would have heuristic value in predicting
antenatal attachment and clinical value in identifying those women at risk for postpartum
84
depression. The knowledge of significant risk in the hospital environment was expected to
interact with these other factors by increasing the level of attachment to the fetus that
hospitalized women report.
CHARACTERISTICS OF THE SAMPLE
Demographic Qualities
In addition to their hospitalization, the 91 women who participated in this research
were quite different from the sample groups of previous studies. Almost half of the
participants were not Caucasian, however the percentage of Caucasian patients who
participated in the research is congruent with the general population of the Dallas area
(53.8% versus 59.1%). The sample percentage of Hispanic patients is slightly smaller (11%
versus 35.6%) and the number of African American patients is slightly higher (33% versus
20.3%) than the Dallas demographic (2004 Dallas Community Census). Previous research
done in this country has been done with samples that were predominantly Caucasian: Studies
utilizing high risk samples like those of Gupton (77.9%) and Maloni (94%) have to this date
underrepresented other ethnicities (Gupton et al., 2001; Maloni, Brezinski-Tomasi, &
Johnson, 2001).
Participants’ average age was 27 years, with a range of 17 to 44. This sample was
slightly younger than other research with high risk populations (Gupton reported a mean age
of 29.27 and Maloni reported 31.2) as well as with samples of women with uncomplicated
pregnancy (Zimmerman and Doan reported 30.06 and Lindgren reported 29.5) (Zimerman &
Doan, 2003; Lindgren, 2001).
85
Socioeconomic status was balanced: 24% reported less than $25, 000 in household
income, 35% reported between $26,000 and $65,000, and 29% exceeded $66,000 annually.
In comparison to Gupton’s sample, this group is somewhat wealthier (Maloni did not report
income). Educational attainment was also well represented. Eleven percent of the sample
did not finish high school, but 25% graduated or received a GED, 33% had some college, and
26% had an undergraduate/graduate degree. Ninety-four percent of Maloni’s sample had
some college (Gupton did not report education), but this is not representative of the Dallas
demographic. According to the 2004 Dallas Community Survey, 76% of Dallas County
residents 25 years and older are High school graduates and 28% have a Bachelor’s degree or
higher. The 91 women included in this research represent the area population fairly well.
According to Cornell University New Service, out-of-wedlock births accounted for
one-third of all U. S. births in 2003, exactly the same ratio of unmarried participants in this
study (Lang, 2005). Sixty-five percent were married or were cohabiting. This also
distinguishes the sample from previous psychosocial studies, as most of the participants in
other samples have been married (Gupton, 85.6%; Maloni, 92.1%).
Pregnancy-Related Qualities
Another distinctive quality of this sample is the range of new mothers to primaparous
mothers. Thirty-two percent of the sample was experiencing a first pregnancy, 24.4% were
experiencing their second, 22% their third, and 21.6% had a range of four to seven previous
pregnancies. Fifty-three percent of Gupton’s participants were pregnant for the first time,
and Lindgren reported her sample as being 41% primaparous. Maloni, in her discussion of
the impact of bed rest upon the families of her participants, reported that 45 of the 89 women
86
had other children. In this sample, 54.9% of the women had at least one child at home (33%
had two children, and 15% had from three to six children in the household).
Psychiatric Qualities
Although psychiatric illness was not a specific interest in this study, it is of interest to
compare the prevalence rates of depression and anxiety published in the Diagnostic and
Statistical Manual of Mental Disorders with the rates found in this sample of women (2000).
Almost seven percent of the sample had history of being diagnosed with depression (point
prevalence in community samples is 5-9%), and the same number reported receiving a
diagnosis of one of the anxiety disorders (one-year prevalence rate in community samples is
5%). Three percent reported a dual diagnosis (community rates are around 10%), and one
participant had previously been diagnosed with bipolar disorder (lifetime prevalence
approximately 0.5%). Almost 42% of the women scored at or over the threshold (score of
11) of the Edinburgh Postpartum Depression Scale (EPDS), and 44% scored at or over the
threshold (score of 16) on the Center for Epidemiologic Studies-Depression scale (CES-D)
provoking the administration of the Structured Clinical Interview for Diagnosis of Axis-I
Disorders (SCID). Yet of these administrations, although two participants fully met the
criteria for a dual diagnosis of Major Depressive Disorder and Generalized Anxiety Disorder,
none met the criteria for a diagnosis of depression alone. Although it is beyond the scope of
this discussion, this is a common finding due to the difficulty of choosing a threshold score
on depression screening instruments with the acceptable balance of specificity and sensitivity
(Austin & Lumley, 2003; Cox & Holden, 2003a). In Austin and Lumley’s review article of
antenatal screening research, four out of sixteen published studies reported similarly high
87
percentages of subjects exceeding the threshold. One possible explanation for the
discrepancy between positive screenings for depression and negative diagnoses in this
sample pertains to the DSM-IV criteria for Major Depressive Disorder. The EPDS instructs
the subject to answer the questions based on how she has felt over the last seven days,
however the DSM-IV criteria require that the depressive symptoms have been in place for
most of the day over the previous two-week period. The participants in this study completed
the questionnaire within the first 72 hours of hospitalization; the clinical interviews were
usually administered at a later point, dependant upon patient and research investigator
availability. Often patients would report that they had been upset initially upon
hospitalization but were “feeling better now that things have settled down.”
Varieties of Obstetric Risk
The Hobel Risk Asssessment system includes 51 antenatal maternal and fetal risk
factors (Tables 28 and 29). Its design was to enable an assessment of prematurity probability
(Hobel et al., 1973). However, the original instrument did not include in the prenatal
inventory four conditions that often present in hospital admission for obstetric risk:
Premature rupture of the membranes, preterm labor, placenta previa, and placental abruption.
In this study, these four common diagnoses were included in the risk assessment, and
weighted for severity by the Principal Investigator of the study, obstetrician John Rosnes.
Three studies provide interesting comparisons of common risks (Table 49). Perhaps
Gupton’s study is the most similar in context, for it included 105 women hospitalized for
more than 48 hours. Maloni’s study sample consisted of 89 women prescribed antepartum
bed rest in the hospital or at home, and participants in Besser’s study were not hospitalized.
88
However, there are significant differences across these three samples, testifying to the
difficulty of comparing this sample with others in the population.
DISCUSSION OF FINDINGS
Maternal Object Representation and Maternal Antenatal Attachment
A previous investigation found significant associations between antenatal attachment
and object representation (Priel et al., 2001), but the data from this sample does not support a
relationship between the two constructs. Multiple statistical analyses, both parametric and
nonparametric, were employed manipulating the sample in several configurations and all
failed to reach significance. Contrary to the hypotheses, it is intriguing that the MANOVA
procedures examining the Maternal Antenatal Attachment Scale (MAAS) scores of Quality,
Intensity, and Global attachment across the Conceptual Levels of the narratives consistently
associated lower attachment scores to the participants rated the highest in developmental
level on the Object Relations Inventory (ORI) narratives. This nonsignificant trend is an
example of the hazard of sampling error, for only eight subjects fell in the lower two levels
of development (Sensorimotor/preoperational and Concrete) in the mother narrative as
opposed to 75 found in the two higher levels of development (External and Internal Iconic).
Although the sample reached the stated minimum of 90 subjects, the power was insufficient
for confidently detecting differences in some analyses, particularly those involving the six
dimensions of the ORI. Nevertheless, this study was an effective pilot in terms of exploring
trends or signals.
89
Priel and Besser employed both the ORI and the MAAS in a study of 120 Israeli
women with uncomplicated pregnancies and a mean age of 25.21 (SD = 3.50; Priel et al.,
2001). Table 50 displays a comparison of the ORI and MAAS means and standard
deviations from their sample with those of this work. The statistics of the two samples
overlap in each score with the exception of “Punitive.” In regard to the attachment measure,
the Quality score means are within one point of one another, while the Baylor sample
Intensity means exceed those of the Israeli sample by three points. Priel and Besser were
able to report significant correlations between each ORI dimension and Quality of
attachment, as well as a significant correlation between Conceptual Level and Intensity of
attachment. Table 51 compares the correlations of the two samples. A primary difference
between the two samples is the presence of obstetric risk in the Baylor subjects, but the
differences in findings are more likely a product of discrepant sample sizes (the Israeli study
included 120 participants, 45% more than the Baylor sample).
In the absence of relationships between the ORI dimensions and the MAAS scales,
there are some interesting relationships within the ORI. One rather strong association found
in the mother narrative is that between the characteristics of Ambivalence and Length (r =
.40, p = .000). The scoring manual of the ORI directs Ambivalence to be evaluated by the
degree to which the subject relates opposing feelings about the “other,” or expressing
confused, mixed feelings. Length is evaluated by counting the lines in the narrative. In
taking a look at the specific narratives with high ambivalence scores, it clearly takes more
fluency for the subject to express the confusion. One narrative demonstrates this vividly:
My mother just turned 44 years old this month. She is also the mother of ten children, but if it weren’t for abortion and miscarriges (sic) she would probably
90
have 25 kids. She had me at the age of 16 and I was raised by my grandmother. My mother has never really had a job for a long time but she is very educated. She as attained several different degrees and not once put one to use. I sometimes believe that she has multiple personalities but she does have a good heart and you just have to know her to love her.
Ambivalence is also strongly negatively related to Benevolence (r = -.41, p = .000) and has
an even stronger positive correlation with the Punitive descriptive (r = .66, p = .000), further
explaining the nature of confusing feelings about the other.
Object Representation of Mother and Baby
One adventure of this study has been the exploration of the ORI narrative of the fetus.
The assumption was that women who had the capability of expressing complex highly
developed narratives about their mother would be able to do the same about their unborn
child. This was to be evaluated by a comparison of the Conceptual Levels of the two
narratives. This proved to be difficult in data collection, for few narratives about babies
could meet the criteria for the higher conceptual levels. Describing the fetus in
preoperational symbiotic language, concrete literal terms, or by fetal movement and activity
was dominant. Attributing thoughts, feelings, values, or understanding the baby on a wide
range of levels was less often noted. (See Table 9 for the frequencies of Conceptual Levels
in both narratives.) There was no significant association between this characteristic of the
two narratives. However, there were associations in the other dimensions of the scale. The
Conceptual Level of the baby narrative was associated with the length of the mother
narrative. Additionally, the length of the baby narrative was positively correlated with length
of the mother narrative (r = .720, p = .000), and ambivalence in both narratives. As pointed
out in the previous section, some scoring requires the subject to write more about the person,
91
so these relationships of length may be more functional than psychologically revealing. The
ORI has been validated in a number of populations and has no reports of level of education
biasing the narratives, but an analysis of this sample was performed for confirmation. A two-
way contingency table analysis failed to find significant differences in either the baby or
mother narratives according to educational level.
Another consideration is whether the baby narrative is actually capturing the mother’s
internal representation of her baby or her ability to imagine her baby. This would not be
contradictory to the theoretical stance that object representations differ from internal working
models in the respect that they introduce wish and fantasy into the mental representation of
the relationship. The fantasies of pregnant women have been alluded to in other antenatal
research; in fact, the absence of antenatal fantasy has been considered dysfunctional (Cohen,
1979; Grace, 1989; Cranley, 1981; Leifer, 1977). In one investigation, 184 pregnant women
were asked, “Please write a few sentences about what you expect your baby to be like”
(Sorenson & Schuelke, 1999). The responses were analyzed by a series of iterations that
ultimately identified ten major themes: Appearance, psychological traits, gender, behavior,
normalization, deification, role relations, impact on parents, spiritual, and ambiguity.
Findings indicated that fantasies tended to develop across gestational age and differed
between multiparas and primigravidas, with multiparous women seeming to be significantly
influenced by their older child. In investigation of this sample, however, no differences in
Conceptual Level of the ORI were found between women who had children at home and
first-time mothers.
92
Dependency, Self-Criticism, and Antenatal Attachment
In studies of romantic attachment in adult samples, dependency has been associated
with an anxious-ambivalent style and self-criticism with a fearful, avoidant style (Besser et
al., 2003a) (Zuroff et al., 1995). Similar findings resulted from research conducted with a
sample of women in uncomplicated pregnancy (Priel et al., 1999). Therefore, it was
expected that dependency in this sample would be related to Condon’s anxious ambivalent
antenatal attachment style (low quality, high intensity) and that self-criticism would be
associated to the avoidant antenatal attachment style (high quality, low intensity; Condon,
1993). A larger sample may have produced significance; however, the trend suggests that the
hypothesis would be disproved even in a larger sample. Those scoring over the mean of the
sample for dependency were fairly equally distributed throughout the four attachment styles.
Out of 39 individuals who scored above the mean for self-criticism, only three resided in the
Avoidant quadrant, whereas the other 36 participants were fairly evenly distributed across the
remaining three antenatal attachment styles.
A question that appears in this analysis is the suitability of the MAAS for this type of
exploration. First, it is possible that the Quality and Intensity scales of the MAAS are not
sufficiently orthogonal to produce the four theorized styles. Second, some of Condon’s
predictions regarding the scale are not confirmed in the sample. For example, he proposed
that multiparous women may be overrepresented in the second quadrant, Avoidant, due to a
lack of time in “attachment mode.” Analyses in this study found no differences between
multiparous and nulliparous women.
93
Self-Criticism, Dependency, and Object Representation
It was predicted that women with a self-critical style would convey maternal object
representations of a higher conceptual level and report a higher quality of antenatal
attachment than women with a dependent style. Only eight subjects responded with
narratives in the lower two conceptual levels opposed to 75 who were rated in the higher two
conceptual levels. Although the differences were not significant, it is noteworthy that
dependent and self-critical styles in the two-way contingency tables (Tables 26 and 27) are
exactly opposite in their placement. Those who are less self-critical tend to provide
narratives of higher conceptual levels, whereas those who are more dependent express
maternal representations of a higher level.
Also, the self-critical type failed to report a higher quality of attachment than the
dependent type. Self-criticism was, in fact, significantly negatively associated with Quality
of attachment (r = -.37, p <.000). This result is interesting in relation to Priel and Besser’s
finding that highly self-critical subjects’ risk for depression was lowered if they became
strongly attached to the fetus during pregnancy. If self-criticism is associated with a lower
quality of antenatal attachment, but attachment can reduce vulnerability to postpartum
depression, an interesting paradox exists.
Risk and Attachment
Previous research has been equivocal concerning the impact of risk on maternal fetal
attachment. Findings from this sample are not ambiguous—they are simply negative. A
number of statistical analyses investigating both type of risk (maternal, fetal, or combined)
and level of risk (low, medium, and high, calculated by the mean of the Hobel Risk
94
Assessment scores of the sample) failed to find any significant relationships between risk,
attachment, and object representation in this sample. Ruling out measurement error, it is
possible that the level or type of risk is not strong enough to affect differences in antenatal
attachment in the women in this sample (Cannella, 2005).
Depression
Both screening measures, the CES-D and the EPDS, were significantly correlated
with the Quality scale of the MAAS. However, only the EPDS was significantly associated
with the Global score, and neither instrument was associated with the Intensity scale. It
would be interesting in future research to investigate any potential associations between the
Intensity of Attachment factor and screening instruments designed for the spectrum of
anxiety disorders. Both screening instruments were significantly correlated with the
Dependency and Self-Criticism scales of the DEQ. However, neither instrument was
associated with the severity of risk.
Exploring the attachment styles of mothers who scored above and below the threshold
for possible depression on the EPDS depression instrument exposed an interesting and
significant relationship between depressive symptomatology and attachment quality. Table
41 portrays mothers over the threshold score of 11 as residing in the third (withdrawn) and
fourth (anxious ambivalent) quadrants of Condon’s model. This is consistent with previous
research referring to the tendency for depressed mothers to be withdrawn from their infants,
at great cost to the child (Dawson, Klinger, Panagiotides, Hill, & Spieker, 1992; Murray,
1992).
95
Alternate Scoring Methods for the DEQ
As mentioned in the results, psychometricians have concerns regarding the standard
scoring system utilized for the DEQ (Santor et al., 1997a; Bagby et al., 1994). In addition, a
growing movement within the domain of feminist psychology has questioned the assumption
of dependency as a maladaptive style of interpersonal relatedness (Rude et al., 1995).
Several theorists, Blatt and the fellow authors of the DEQ included, have suggested that the
Dependency scale might actually be composed of two subscales, a “healthy” one and an
“unhealthy” one. This has led to revisions of the scoring system and new subscales of
“Relatedness” and “Dependency” (Blatt et al., 1995; Bacchiochi et al., 2003) and
“Connectedness and Neediness” (Rude et al., 1995). Advocates of this stance believe this
distinction between the psychologically adaptive maintenance of close, reciprocal
relationships and the pathological fearful, helpless, and clinging approach to others may
explain why often those who score highly in Dependency are less vulnerable to depression
1995). As seen in Table 43, the samples’ DEQ scores were calculated by each method and
then compared for associations. Of note are the positive correlations between the two Self-
Criticism scores and Rude and Burnham’s Neediness subscale, and the lack of association
between the Self-Criticism scores and the Connectedness subscale. The difference, while
still significant, is not so striking when comparing Blatt’s subscales, Dependency and
Relatedness, with the standard scales. This generates the question of what is being measured,
personality tendencies or a vulnerability to depression that transcends self-criticism or
dependency?
96
When testing the other scoring methods for associations with attachment, there are
significant findings. The McGill scoring system produces the same negative correlation
between Self-Criticism and Quality of attachment (r = -.339, p < .001), as the standard scale
(r = -.336, p <.000), and correlations between Dependency, Intensity of attachment, and
Global attachment are all nonsignificant. Blatt et al’s subscale Relatedness (the healthy,
mature form of dependency) has a significant negative association with the Quality of
Attachment factor (r = -.24, p = .02). Adding to the confusion, Rude and Burnham’s
Neediness (the unhealthy, pathological type of dependency), has the exact same significant
negative association with Quality of attachment (r =-.24, p =.02). It is difficult to
understand why both the healthy and unhealthy forms of dependency would be negatively
correlated with attachment quality. This might be instrument error, or some piece of the
Dependency construct might be relating in an underlying fashion to the attachment measure.
When all eight scores are subjected to a linear regression with the Global attachment
score, all fail to reach significance. The standard scoring system and the McGill scoring
system are slightly more powerful than the subscales of the Dependency factor, no doubt
because of the inclusion of the Self-Criticism factor. Although it is possible that a larger
sample size would drive some of these analyses into significance, it appears that the
relationship of self-critical and dependent tendencies with attachment is far less important
than the relationship between depression and attachment. The Self-Critical and Dependent
factors of the DEQ may be more valuable in identifying risk factors for postpartum
depression than in identifying risk factors for low or poor quality attachment.
97
Gestational Age and Attachment
Since previous research has suggested that attachment grows with gestational age, it
was surprising that gestational age was not associated with the level of attachment reported
by the participants. A common observation across studies has been that fetal movement is a
trigger for increased antenatal attachment, and all subjects in this sample had experienced
quickening (Mikhail et al., 1991; Reading et al., 1984). This may have rendered the analyses
useless. However, one narrative underscores the individual nature of antenatal bonding:
“…I must say that he wasn’t planned. Me and his father were not trying to get pregnant, but it happened so quickly my eyes didn’t have time to blink.
I didn’t bond with him until I was almost 5 months pregnant. At this time, that’s when the secret about me being pregnant was finally out of the bag to my family and everybody I hid it from…”
Another confounding issue with this research is that all participants had received at
least one sonogram, many of them three-dimensional, and had pictures of the fetus at
bedside. The impact upon antenatal attachment of viewing the fetus through sonogram has
been under investigation although findings have been inconclusive (Reading et al., 1984;
Righetti, Dell'Avanzo, Grigio, & Nicolini, 2005).
Exploring Fetal Object Representation
In addition to testing the hypotheses, a few other questions were asked. Joy Penticuff
highlighted the developmental resolution of ambivalence toward a new member of the family
that occurs over the course of pregnancy (Penticuff, 1982). It was supposed that this
ambivalence was intensified in mothers with higher obstetric risks. However, in all the
analyses of risk, no significant association was found between the Ambivalence dimension of
the ORI and risk. There was a significant association between Ambivalence in the mother
98
narrative and Ambivalence in the baby narrative, suggesting that these conflicting feelings
may be more a characteristic of the mother rather than of a particular mental representation.
In addition, Ambivalence was negatively correlated with the Global attachment score,
implying that this characteristic has implications for antenatal attachment.
Other analyses tested for relationships between the ORI dimensions, ethnicity and
age, with no significant differences found in the narratives. The ability to convey complex
information about mother or baby does not seem to be related to factors outside the process
of mental representation.
A final curiosity about the fetal narrative had to do with the theories behind internal
working models and mental representations. If these are activated during infancy and are
primarily associated with caregiver relationships (and later to romantic attachment figures),
perhaps there is a distinctly different model or representation a mother constructs of her
child. This model and representation may be focused on caregiving, instead of care
“receiving.” If so, multiparous women might have a certain facility with the mental
representation of their baby that nulliparous women, with no previous model or
representation of a baby, do not possess. This was not borne out statistically and, although
the sample size was adequate for the chi-square analysis, there were no significant
differences in the Conceptual Level of the fetal narratives according to parity.
99
Maternal Antenatal Attachment Scale
The positive correlation between age and the Quality of attachment factor was
moderately strong (r = .292, p = .005). It appears that older mothers may have more positive
thoughts about their babies than younger ones, although there is no difference in their level of
preoccupation with the fetus. It has also been stated that multiparous women spend less time
in “attachment mode” because of caregiving responsibilities for other children. However,
this was unsupported in the sample, as there were no differences in attachment scores
between first-time mothers and those with children at home.
THEORETICAL IMPLICATIONS
While this study did not find the proposed relationships between personality
characteristics, mental representation, and attachment, the limitations of the sample size warn
against premature conclusions. This work did, however, contribute to the ongoing dialogue
regarding the construct of antenatal attachment and its associations. In addition, it provided a
valuable, diverse sample of hospitalized women who enabled research to ask old questions in
a new context.
CLINICAL IMPLICATIONS
Even though the construct of antenatal attachment remains elusive, this research also
adds value to the growing investigation into the insidious link between depression and
attachment. Whether depressed mothers suffer from impaired attachment or a mother’s
inability to attach to a baby contributes to a vulnerability to depression remains to be seen.
100
These findings support all those before that point to the necessity of identifying, targeting,
and intervening where possible to enhance maternal-fetal attachment and treat perinatal
depression.
The idea that self-critical tendencies are a detractor of attachment as well as a strong
contributor to depressive symptomatology is not new, but this research has broadened its
application to the population of women with high-risk pregnancy. In fact, self-criticism may
be a salient contributor in this population in view of the tendency for women with high-risk
to look inward for a causal relationship between something they did or did not do and the
complications. The findings in this research, consistent with those of the Priel and Besser
team in Israel, suggest that pregnant women who are self-critical are vulnerable to depressive
feelings and may benefit from interventions that expose these tendencies. In addition, since
self-criticism also impacts antenatal attachment, interventions that enhance attachment to the
fetus may be indicated.
Although cognitive-behavioral and interpersonal approaches with postpartum
depression have been empirically studied (Dennis, 2004), only one reference was found that
suggests treating childbearing depression from an attachment theory framework (Whiffen &
Johnson, 2006). This case example illustrated how Emotionally Focused Marital Therapy, a
manualized psychotherapy based upon attachment theory, could be employed in treating
postpartum depression. Social support has been found to be protective against perinatal
depression (Priel et al., 2000a), so targeting depression from a “couples” standpoint would be
ideal for enhancing attachment and encouraging partner support. This could be even more
critical in women with obstetric risks. Hospitalization separates them from their partners,
101
increases the partner burden in terms of household, family, and financial responsibilities, and
sets in place a fertile environment for partner blame and discord. Progressive hospitals have
added group therapy to antepartum unit care for psychosocial support with equivocal results
(Dennis, 2004). Perhaps focusing on the marital dyad would be more effective in the
prevention of perinatal depression. Interventions that highlight emotional expression and
foster understanding of each spouse’s needs and childbearing fears could be operationalized
in group sessions, similar to childbirth classes, or in the hospital room privately.
LIMITATIONS AND FUTURE DIRECTIONS
The study shared two common limitations to research of this sort. Self-report
measures are often the only practical way to approach cross-sectional research. In the battery
of measures for this study, other than the ORI, the questionnaires were all forced choice
Likert-type instruments. Future research might include other forms of data gathering, such as
structured interviews or family reports. In addition, although the instruments utilized in this
study are considered to be reliable and valid, more research is needed to assess the extent to
which they genuinely evaluate these particular constructs of personality, mental
representation, and attachment.
A possible limitation for this work specifically has to do with the method of
administration of the ORI. Because this investigation was a small piece of a larger study
including a number of measures, it was decided to include the ORI in the questionnaire
packet. Although study personnel conscientiously explained to the participants to “take five
minutes and write a description of your mother and the baby you are carrying,” the packets
102
were left for completion and in some cases the narratives were either skipped (five of the 91
subjects did not complete either the mother or the baby narrative and an additional three did
not complete the mother narrative) or so short that some of the descriptive qualities were by
necessity scored as “missing.” Future studies of this sort might consider conducting the ORI
in more of an interview fashion. Initial remarks could be recorded and probing questions
asked if additional information is needed.
Rating the fetal narratives according to the standard scoring system was not possible,
as explained earlier, because the descriptives were inappropriate in application to a fetus
An asterisk indicates tht the 95% confidence interval does not contain zero, and therefore the difference in means is significant at the .95 significance using Dunnett’s C procedure.
164
Table 41 Chi-Square Comparison of EPDS Depressive Symptomatology and MAAS Antenatal Attachment Style
EPDS Below Threshold
EPDS
Above Threshold
Antenatal Attachment Style % (N) % (N) 1) High Quality, High Intensity (Strong, secure)
26.4 (24)
6.6 (6)
χ² = 21.339, p = .000
2) High Quality, Low Intensity (Anxious Avoidant)
15.4 (14)
3.3 (3)
3) Low Quality, Low Intensity (Withdrawn)
7.7 (7)
19.8 (18)
4) Low Quality, High Intensity (Anxious Ambivalent)
Table 47 Linear Regression Analyses of Dependency and Self-Criticism Scores Predicting MAAS Global Attachment Score
95% CI
F (2, 88)
p
R²
Adj. R²
Standard Scoring
Dependency -1.32 to 1.96 Self-Criticism
-2.86 to .064
1.87
.16
.04
.019
McGill Scoring
Dependency -.08 to .08 Self-Criticism
-.15 to .002
1.86
.16
.04
.019
Blatt Subscales
Dependency -.37 to .13 Relatedness
-.21 to .22
.78
.46
.02
-.005
Rude & Burnham
Neediness -2.94 to 1.03 Connected
-1.44 to 2.01
.457
.63
.01
-.01
Note: N = 89
171
Table 48 Spearman’s rho Correlation of ORI Baby Narrative and Gestational Age (Weeks)
ORI Baby Narrative
Conceptual Level Length Ambivalence Conceptual Level X X X Length .127 X X Ambivalence -.170 .355** X Gestational Age (Weeks) -.114 .060 -.038
Note: N = 85
ORI = Object Relations Inventory ** = p = .001
172
Table 49 Comparison of Most Common Risk Factors
Baylor Sample Maloni Sample
Besser Sample
Gupton Sample*
N % N
% N
%
N %
Preterm Labor
40
43.95 41 46.06 X X X 20
Placenta Previa
3 3.29 7 7.86 X X X 20
Incompetent Cervix
29 31.87 6 6.74 X X X X
Cervical Abnormality
2 2.19 5 5.61 X X X X
Pregnancy-Induced Hypertension
18 19.78 3 3.3 X X X 18
Premature Rupture of Membranes
16 17.58 2 2.2 X X X 17
Other X X X 28.23 X X X 25
Diabetes
5 5.49 X X 146 100% X X
Total
91 ** 89 100 146 100% 105 100
Note: *Gupton did not include exact N per condition ** Percentages exceed 100% in view of dual or multiple diagnoses (Maloni et al., 2001; Besser et al., 2002; Gupton et al., 2001)
Table 50 Comparison of the ORI and MAAS Means and Standard Deviations from Two Samples
Baylor Sample
Israeli Sample
M SD M
SD
Object Representations (Mother)
Benevolent
4.11 1.54 4.62 1.20
Punitive
1.49 1.42 3.97 1.07
Ambitious
3.53 1.69 3.90 1.18
Ambivalent
1.83 1.30 2.61 1.32
Conceptual Level
5.17 1.22 5.34 2.08
Antenatal Attachment
Quality
46.01 3.48 45.21 4.24
Intensity
30.71 4.88 27.72 4.96
Note: Baylor Sample: Object Relations Inventory (ORI) N = 83
Maternal Antenatal Attachment Scale (MAAS) N = 91 Israeli Sample: N = 120 (Priel et al., 2001)
173
Table 51 Comparisons of Correlations for ORI Mother Narrative and Maternal Antenatal Attachment in Two Samples Baylor Sample
N = 83 Israeli Sample
N = 120 Quality Intensity Quality Intensity Object Representations (Mother)
CENTER FOR EPIDEMIOLOGIC STUDIES—DEPRESSION SCALE (CES-D) Below is a list of some ways you may have felt or behaved. Please indicate how often you have felt this way during the last week by checking the appropriate space. Durin r
the
of a
days)
M
Time (5-7
1. I was bothered by things that usually don’t bother me.
0 1 3
2. I d ing; my appetite was poor.
0 1 2 3
3. I felt that I could not shake off the blues even with
0 1 2 3
4. I felt I was just as good as other people.
0 1 2 3
5. I had as do
0 1
6. I felt depressed.
0 1 2 3
7. I felt that everything I did was an effort.
0 1 2 3
8. I felt hopeful about the future.
0 1 2 3
9. I thou t my life had been a failure.
0 1 2 3
10. I f
0 1 2 3
11. My sleep was restless.
0 1 2 3
12. I was happy.
0 1 3
13. I talke
0 1 3
14. I f
0 1 2 3
15. People were unfriendly.
0 1 2 3
16. I enjoyed life.
0 1 2 3
17 I had
0 1
18. I felt sad. 0 1 2 3
19 felt liked me.
0 1
20. I could not get going. 0 1 2 3
g the past week: Rarely onone oftime (less than 1 day)
Some or alittlethe time (1-2 days)
Occasionally or
Moderate amount ofTime (3-4
ost or all of the
days)
2
id not feel like eat
help from my family or friends.
trouble keeping my mind on what I wing.
2 3
gh
elt fearful.
2
d less than usual.
elt lonely.
2
. crying spells. 2 3
. I that people dis 2 3
178
DEPRESSIVE EXPERIENCE UESTIONN E (DEQ)
Listed below er of statements concerning personal characteristics and traits. Read each item and decide u agree or disagree and to what extent. If you strongly agree
S Q AIR
are a numbwhether yo , circle7; if you
strongly disagree, circle 1; The midpoint, if you are neutral or undecided, is 4.
Strongly Strongly Disagree Agree
1. I set my personal goals and standards as high
as possible. 1 2 3 4 5 6 7
2. Without s pport from others who are close to me, I would be helpless. 1 2 3 4 5 6 7
3. I tend to isfied with my current plans and goals,
rat th 1 2 3 4 5 6 7
4. Sometimes I feel very big, and other times I feel v small. 2 3 5 6 7
5. When I am closely involved with someone, I never feel jealous. 1 2 3 4 5 6 7
6. I urgently need things that only other people
can provide. 1 2 3 4 5 6 7
7. I often find that I don't live up to my own standards or ideals. 1 2 3 4 5 6 7
8. I feel I am always making full use of my potential
abilities. 1 2 3 4 5 6 7 9. The lack of permanence in human relationships
doesn't bother me. 1 2 3 4 5 6 7 10. If I fail to live up to expectations, I feel unworthy. 1 2 3 4 5 6 7 11. Many times I feel helpless. 1 2 3 4 5 6 7 12. I seldom worry about being criticized for things
I have said or done. 1 2 3 4 5 6 7
13. There is a considerable difference between how I am now and how I would like to be. 1 2 3 4 5 6 7
14. I enjoy sharp competition with others. 1 2 3 4 5 6 7 15. I feel I have many responsibilities that I must meet. 1 2 3 4 5 6 7 16. There are times when I feel "empty" inside. 1 2 3 4 5 6 7 17. I tend not to be satisfied with what I have. 1 2 3 4 5 6 7 Copyright: Sidney J. Blatt, Ph.D., Joseph P. D'Afflitti, Ph.D., Donald M. Quinlan, Ph.D., 1979.
u
be sather an striving for higher goals.
ery 1 4
179
19. I become frightened when I feel alone. 1 2 3 4 5 6 7 20. I would feel like I'd be losing an important part
of myself if I lost a very close friend. 1 2 3 4 5 6 7
2 1 2 3 4 5 6 7
22. I have difficulty breaking off a relationship
1 2 3 4 5 6 7 23. I often think about the danger of losing someone
1 2 3 4 5 6 7
25.
1 2 3 4 5 6 7 26. I am not very concerned with how other people
1 2 3 4 5 6 7 27.
ejection. 1 2 3 4 5 6 7
29. It's important for my family that I succeed. 1 2 3 4 5 6 7
30. Often, I feel I have disappointed others. 1 2 3 4 5 6 7
lo . 1 2 3 4 5 6 7
t )
sts friend 1 2 3 4 5 6 7
lati nship. 1 2 3 4 5 6 7
are times when I feel extremely good about myself and other times
3 4 5 6 7
18. I don't care whether or not I live up to what other people expect of me. 1 2 3 4 5 6 7
1. People will accept me no matter how many mistakes
I have made.
that is making me unhappy.
who is close to me. 1 2 3 4 5 6 7
24. Other people have high expectations of me.
When I am with others, I tend to devalue or "undersell" myself.
respond to me.
No matter how close a relationship between two people is, there is always a large amount of uncertainty and conflict. 1 2 3 4 5 6 7
28. I am very sensitive to others for signs of r
31. If someone makes me angry, I let him (her) know
how I feel. 1 2 3 4 5 6 7
32. I constantly try, and very often go out of my way, to please or help people I am c se to
33. I have many inner resources (abilities, s rengths . 1 2 3 4 5 6 7
34. I find it very difficult to say "No" to the reque of s. 35. I never really feel secure in a close re o
. The way I feel about myself frequently varies: there36
when I see only the bad in me and feel like a total failure 1 2
180
7. Often, I feel threatened by change. 1 2 3 4 5 6 7
o
leave, I could still "go it alone." 1 2 3 4 5 6 7
her
.
gs e le
1 2 3 4 5 6 7 3. I often feel guilty. 1 2 3 4 5 6 7
4. I think of myself as a very complex person, one 1 2 3 4 5 6 7
5. I worry a lot about offending or hurting someone
8. I feel good about myself whether I succeed or fail. 1 2 3 4 5 6 7
9. I can easily put my own feelings and problems aside,
problems of someone else. 1 2 3 4 5 6 7
50.would feel threatened that he (she) might leave me. 1 2 3 4 5 6 7
51. rtant responsibilities. 1 2 3 4 5 6 7
2. After a fight with a friend, I must make amends as
s .
1 2 3 4 5 6 7
3
38. Even if the person who is closest t me were to
39. One must continually work to gain love from anotperson: that is, love has to be earned 1 2 3 4 5 6 7
40. I am very sensitive to the effects my words or
actions have on the feelin of oth r peop . 1 2 3 4 5 6 7 41. I often blame myself for things I have done or
said to someone. 1 2 3 4 5 6 7 42. I am a very independent person.
4 4
who has "many sides."
4who is close to me.
4
It is not "who you are," but "what you have
4 4
and devote my complete attention to the feelings and
If someone I cared about became angry with me, I
I feel comfortable when I am given impo
5
soon as possible. 1 2 3 4 5 6 7 53. I have a difficult time accepting weaknesses in my elf 1 2 3 4 5 6 7 54. It is more important that I enjoy my work than it
is for me to have my work approved.
181
55. After an argument, I feel very lonely. 1 2 3 4 5 6 7
56. In my relationships with others, I am very concerned 1 2 3 4 5 6 7
57. rarely think about my family. 1 2 3 4 5 6 7
me vary:
when I feel all-loving towards that person. 1 2 3 4 5 6 7
59. those around me. 1 2 3 4 5 6 7
60. " l.
ys m
1 2 3 4 5 6 7 4. I tend to be very critical of myself. 1 2 3 4 5 6 7
65. l
ar o
about what they can give to me.
I
58. Very frequently, my feelings toward someone close tothere are times when I feel completely angry and other times
What I do and say has a very strong impact on
I sometimes feel that I am specia " 1 2 3 4 5 6 7
61. I grew up in an extremely close family. 1 2 3 4 5 6 7 62. I am very satisfied with m elf and my accomplish ents. 1 2 3 4 5 6 7 63. I want many things from someone I am close to.
6
Being alone doesn't bother me at a l. 1 2 3 4 5 6 7 66. I very frequently compare myself to stand ds or g als. 1 2 3 4 5 6 7
182
EDINBURGH POSTPARTUM DEPRESSION SCALE (EPDS)
Please circle the answ how you have felt over the past 7 days. er that best describes In the past 7 days: 1. ny sid of thin - I have been able to laugh and see the fun e gs 0 ys As much as I alwa could 1 Not quite so much now 2 Definitely not so much now 3 Not at all
2. I have looked forward with enjoyment to things - 0 As much as I ever did 1 Rather less than I used to 2 Definitely less than I used to 3 Hardly at all 3. I ha ent wrong - ve blamed myself unnecessarily when things w 0 No, not at all 1 Hardly ever 2 Yes, sometimes 3 Yes, very often 4. I have been anxious or worried for no good reason - 3 Yes, quite a lot 2 Yes, sometimes 1 No, not much 0 No, not at all 5. I have felt scared or panicky for no very good reason - 3 Yes, quite a lot 2 Yes, sometimes 1 No, not much 0 No, not at all
183
6. Things have been getting on top of me - 3 Yes, most of the time I haven't been able to cope at all 2 Yes, sometimes I haven't been coping as well as usual 1 No, most of the time I have coped quite well 0 No, I have been coping as well as ever
7. I have been so unhappy that I have had difficulty sleeping - 3 Yes, most of the time 2 Yes, sometimes 1 Not very often 0 No, not at all
8. I have felt sad or m erable - is 3 Yes, most of the time 2 Yes, quite often 1 Not very often 0 No, not at all
9. I have been so unhappy that I have been crying - 3 Yes, most of the time 2 Yes, quite often 1 Only occasionally 0 No, never
10. The thought of harming myself has occurred to me - 3 Yes, quite often 2 Sometimes 1 Hardly ever 0 Never
(J.L. Cox, J.M. Holden, R. Sagovsky, Department of Psychiatry, University of Edinburgh)
184
MATERNAL ANTENATAL ATTACHMENT SCALE (MAAS)
These questions are about your thoughts and feelings about the developing baby. Please tick one box only in answer to each question. 1) Over the past two weeks I have thought about, or been preoccupied with the baby inside me:
Almost all the time
Very fre uq ently
Frequently
Occasionally
Not at all 2) Over the past two weeks when I have spoken about, or thought about the baby inside me I got
emotional feelings which were:
Very weak or non-existent
Fairly weak
In between strong and weak
Fairly strong
Very strong 3) Over the past two weeks my feelings about the baby inside me have been:
Very positive
Mainly positive
Mixed positive and negative
Mainly negative
Very negative
185
4) Over the past two weeks I have had the desire to read about or get information about the developing baby. This desire is:
Very weak or non-existent
Fairly weak
Neither strong nor weak
Moderately strong
Very strong
5) O r tve he past two weeks I have been trying to picture in my mind what the developing baby
actually looks like in my womb:
Almost all the time
Very frequently
Frequently
Occasionally
Not at all 6) Over the past two weeks I think of the developing baby mostly as:
A real little person with special characteristics
A baby like any other baby
A human being
A living thing
A thing not yet really alive
186
7) Over the past two weeks I have felt that the baby inside me is dependent on me for its well-being:
Totally
A great deal
Moderately
Slightly
Not at all 8) Over the past two weeks I have found myself talking to my baby when I am alone:
Not at all
Occasionally
Frequently
Very frequently
Almost all the time I am alone 9) Over the past two weeks when I think about (or talk to) my baby inside me, my thoughts:
Are always tender and loving
Are mostly tender and loving
Are a mixture of both tenderness and irritation
Contain a fair bit of irritation
Contain a lot of irritation
187
10) The picture in my mind of what the baby at this stage actually looks like inside the womb is:
Very clear
Fairly clear
Fairly vague
Very vague
I have no idea at all 11) Over the past two weeks when I think about the baby inside me I get feelings which are:
Very sad
Moderately sad
A mixture of happiness and sadness
Moderately happy
Very happy 12) Some pregnant women sometimes get so irritated by the baby inside them that they feel like they want to hurt it or punish it:
I couldn’t imagine I would ever feel like this
I could imagine I might sometimes feel like this, but I never actually have
I have felt like this once or twice myself
I have occasionally felt like this myself
I have often felt like this myself
188
13) Over the past two weeks I have felt:
Very emotionally distant from my baby
Moderately emotionally distant from my baby
Not particularly emotionally close to my baby
Moderately close emotionally to my baby
Very close emotionally to my baby 14) Over the past two weeks I have taken care with what I eat to make sure the baby gets a good diet:
Not at all
Once or twice when I ate
Occasionally when I ate
Quite often when I ate
Every time I ate 15) When I first see my baby after the birth I expect I will feel:
Intense affection
Mostly affection
Dislike about one or two aspects of the baby
Dislike about quite a few aspects of the baby
Mostly dislike
189
16) When my baby is born I would like to hold the baby:
Immediately
After it has been wrapped in a blanket
After it has been washed
After a few hours for things to settle down
The next day 17) Over the past two weeks I have had dreams about the pregnancy or baby:
Not at all
Occasionally
Frequently
Very frequently
Almost every night 18) Over the past two weeks I have found myself feeling, or rubbing with my hand, the outside of my stomach where the baby is:
A lot of times each day
At least once per day
Occasionally
Once only
Not at all
190
19) If the pregnancy was lost at this time (due to miscarriage or other accidental event) without any pain or injury to myself, I expect I would feel:
Very pleased
Moderately pleased
Neutral (i.e. neither sad nor pleased, or mixed feelings)
Moderately sad
Very sad
Copywrite JT Condon Dept. Psychiatry Flinders Medical Centre, South Australia
191
Chart Review Participant number Obstetrician of record Birthdate Ethnicity Af Am Asian Caucasian Latino Other Marital Status Single Married Separated Divorced Widowed CohabitingTotal Pregnancies (Prior) Full Term (Prior) Premature (Prior) Abortions induced Abortions spontaneous Ectopics Multiple births (Prior) Living Stillborn List any interventions that have been initiated or ordered by the doctor: HOBEL RISK ASSESSMENT--PRENATAL yes no Moderate to severe toxemia 10 0 Chronic Hypertension 10 0 Moderate to severe renal disease 10 0 Severe heart disease, Class II-IV 10 0 History of eclampsia 5 0 History of pyelitis 5 0 Class I heart disease 5 0 Mild toxemia 5 0 Acute pyelonephritis 5 0 History of cystitis 1 0 Acute cystitis 1 0 History of toxemia 1 0 Diabetes > Class A-II 10 0 Previous endocrine ablation 10 0 Thyroid disease 5 0 Prediabetes (A-I) 5 0 Family history of diabetes 1 0